Inside a launch decision
GreenLeaf’s UK launch: follow the journey, check the evidence, open the reports.
Fictional company · synthetic data
Explore the launchAd clicks11,760
Landing page10,080
Quiz starts8,400
Completed3,780
First purchase259
September 1–14, 2026 · seven-day cohorts · retrospective snapshot September 22
In this example
Meet GreenLeaf
A fictional US supplement brand entering the UK.
The launch hypothesis
Will ingredient transparency help UK buyers choose GreenLeaf?
- Quiz completion target
- 45%
- Monthly UK revenue target
- $60,000
- Unaided awareness target
- 15%
Read the business context
English editorial source — publication and owner/human release pending. GreenLeaf is fictional. Its campaign data and later context update were authored for teaching; the recorded AI responses are real. The checks below are independent calculations and AI-assisted editorial review, not a completed human review or client approval. This example does not measure AI accuracy, human superiority, time saved or campaign ROI.
GreenLeaf Wellness sells subscription supplements: daily greens, protein blends and adaptogen stacks. It is expanding from the US into the UK with a specific proposition: customers can inspect ingredient sourcing and doses instead of trusting an unexplained proprietary blend.
The launch question is simple to ask and harder to answer: will UK buyers care enough to choose it? To make a useful recommendation, the team must connect the customer hypothesis, campaign choices and measurement plan to the results. This path follows those handoffs into one complete reporting example.
Here is the decision at the center of it. Paid-mobile quiz completion falls from 3,780 of 8,400 starts (45%) to 2,352 of 8,400 (28%). An AI draft recommends a diagnostic review, but also supplies incorrect financial totals and dismisses a meaningful change in audience mix. Checked calculations support investigating both cohort composition and the mobile journey. A later fictional product/support note makes the investigation more specific, while leaving its cause unresolved.
The periods are September 1–7 and September 8–14, 2026, using UTC paid-click cohorts. Funnel events and purchases have the same seven-day observation window. The snapshot is September 22 at 23:59:59 UTC. Both periods are mature: this is a retrospective comparison, not evidence that an alert caught the issue during the launch.
Start with the decision you own
| Your responsibility | What you need from this work |
|---|---|
| Agency leadership | Advice you can defend, consistent facts across deliverables and a clear client release decision. |
| In-house marketing leadership | Evidence for changing the launch priority, and a distinction between diagnosis and commercial approval. |
| Media or performance leadership | Comparable cohorts, event definitions, checked calculations and a specific investigation. |
| Creative, product or implementation work | A testable question, the evidence missing from the current data and an accountable next step. |
You can inspect every deliverable. These reader perspectives share one company and one dataset; they do not manufacture different facts. The course's three report recipients remain the founder, media team and content/creative team.
Understand the customer before interpreting the campaign
The course introduces Oliver, a skeptical biohacker who wants to audit the evidence behind a supplement. He is a synthetic persona hypothesis, not a customer who was interviewed.
A useful handoff to the launch team looks like this:
| Working hypothesis | Question for real research | Consequence for the campaign |
|---|---|---|
| Ingredient transparency matters | What did buyers verify before their last supplement purchase? | Test whether sourcing and dosing proof deserve prominence. |
| A lower price can raise doubts about quality | What made an offer feel credible or suspicious? | Explain value with supporting information rather than an unsupported quality claim. |
| A quiz can help buyers choose | What makes buyers trust or abandon that journey? | Examine the post-click experience as well as the ad. |
The reconstruction uses skeptical and exploring audience buckets and traceability and value messages. These are constructed categories for analysis, not validated psychological segments or randomized assignments. The illustrative traceability message is “Know what is in every serving.” Any real deployment would need substantiated product information and appropriate claim review.
The activation handoff is concrete: ad → landing page → quiz start → quiz completion → first purchase, with consistent campaign, device, audience and creative identifiers. The reporting team needs the naming and event definitions, plus release and test history. An ad's message cannot explain a broken completion event; a plausible persona cannot prove a message works.
Define success before opening the export
| Business objective | Fictional course target | What these inputs can establish |
|---|---|---|
| Help buyers progress through consideration | 45% quiz completion | Completions ÷ starts in the supplied paid cohorts. |
| Generate new UK customer revenue | $60,000 per month | Attributed first-purchase revenue from two paid-click cohort weeks, not the full month's result. |
| Build recognition | 15% unaided awareness | Unavailable; clicks and impressions cannot replace awareness research. |
| Acquire customers with lasting value | $480 LTV at 12 months | Unavailable; first-order revenue cannot establish retention or LTV. |
CTR, CPC and CPM help investigate performance. They do not replace these outcomes. The course also gives a $312 breakeven CAC guardrail, but the export contains only media cost and attributed first-purchase customers. Media CPA is not full CAC. Applying that guardrail requires a like-for-like reconciliation of costs, customers and economic assumptions.
Reporting brief v1
GreenLeaf UK launch — reporting brief v1
Version: 1.0 · Authored: 2026-10-29 · Status: English source input for a fictional teaching example; not a published report. Freeze this brief, the dictionary and the source map before fixture generation. Freeze exact run-input hashes separately after data verification. No AI response or analytical result is claimed by this brief.
Business and the handoffs into reporting
GreenLeaf Wellness is the course's fictional subscription supplement business, expanding from the US into the UK. Its products include daily greens, protein blends and adaptogen stacks. Its proposed distinction is ingredient traceability and transparent dosing. The commercial question is whether that proposition helps UK buyers progress from consideration to a first purchase. This reconstruction examines paid cohorts; it cannot settle the whole launch's success.
| Upstream work | Concrete handoff supplied here | How reporting uses it | Evidence still needed |
|---|---|---|---|
| Research and strategy | Oliver, the skeptical biohacker, is a synthetic persona hypothesis. Objection: can I verify what is in a serving, and does the price reflect value? Interview question: what evidence changed your last supplement shortlist? | Compare the skeptical and exploring audience hypotheses and the traceability and value messages without treating identifiers as validated psychology. | Real customer interviews, observed objections and a validated segment definition. |
| Measurement | Business objective → KPI → diagnostic lever. Quiz completion = completions / starts; new-customer revenue has a monthly target. Seven-day UTC paid-click cohorts have a common maturity rule. | Keep outcomes above CTR/CPC/CPM and compare like-for-like denominators. | All-channel monthly revenue reconciliation, awareness research, retention history and incrementality design/results. |
| Activation | Illustrative message: “Know what is in every serving.” Second message tests value through clear dosing information. Journey: ad → landing page → quiz start → completion → first purchase. Canonical campaign/device/audience/creative IDs are in the dictionary. | Inspect message, audience, device and funnel stage; request event and release reconciliation before diagnosing. | Substantiation of any product claim, event QA logs, ad-platform exports and a timestamped release/test register. These illustrative messages are not approved advertising. |
| Institutional memory | This v1 records all scenario context currently supplied. There is no supplied historical seasonal baseline, client preference log, earlier agency finding or known product incident. | Treat missing history as unknown; ask its owner rather than invent a founder sensitivity or seasonal explanation. | Product/support and commercial-owner follow-up; record subsequently learned information with provenance in v2. |
The source map separates course facts, new reconstruction choices and later derived results. Source: course traceability. Field semantics: dictionary.
1. KPIs and targets
| Business decision | Course scenario target | Evidence available from this reconstruction | Reporting boundary |
|---|---|---|---|
| Are buyers progressing through consideration? | 45% quiz completion | Paid-cohort quiz completions ÷ starts, including mobile and desktop separately | The authored mobile comparison is 45% → 28%, a 17 percentage point decline (approximately 37.8% relative). Verify counts after generation; this is a teaching constraint, not an observed client result or all-site rate. |
| Is the UK launch producing new-customer revenue? | USD $60,000 per month | Attributed first-purchase revenue for September 1–14 paid-click cohorts | Two paid-cohort weeks do not establish a full-month, all-channel result. Do not prorate the target into a success verdict. Obtain the full month, common revenue definition and reconciled new-customer ledger. |
| Is GreenLeaf becoming known? | 15% unaided awareness | Unavailable | Impressions, clicks and paid revenue cannot substitute for a brand-awareness study. |
| Are customers valuable over time? | USD $480 LTV at 12 months | Unavailable | First-order revenue cannot establish realized LTV or retention. |
The course's $312 breakeven CAC is a scenario planning guardrail. This export supplies media spend and attributed first-purchase customers, so it supports media CPA, not full CAC. Reconcile acquisition cost scope, unique new customers, periods and the economic assumptions before comparing with that guardrail. No CAC breach or profitability claim follows from media CPA alone.
CTR, CPC, CPM, click-to-start rate, completion-to-purchase rate, media CPA and attributed first-order ROAS are diagnostic levers. Frequency, impression share and bounce rate are unavailable because their required source fields are absent. Incremental lift, all-site revenue, refunds outside the fixture, full acquisition economics and long-term outcomes remain unknown.
2. Stakeholders, decisions and release authority
One prompt must produce all three drafts from the same brief and reconciled dataset. Every finding answers WHAT happened → WHY it matters → WHAT'S NEXT, separating checked facts, hypotheses, unknowns and proposed actions.
| Recipient | Decision the report must support | Required deliverable | Authority and boundary |
|---|---|---|---|
| Founder / CEO (or in-house marketing leader) | Should launch priorities change now, or should the team obtain specific evidence before a commercial decision? | One-page summary of the business signal, material uncertainty and a named next decision. | Founder/marketing owner approves commercial priorities and budget. No automatic reallocation, unsupported lost-revenue estimate or monthly target verdict. |
| Media team | Which cohort, event definition or journey stage should be investigated first, and what evidence would change the next media decision? | Channel/device/audience/creative analysis, funnel counts, weighted rates and fixed-mix check. | Media lead owns source reconciliation and diagnostic work order; spend changes require commercial approval. |
| Content & creative team | Should the next test focus on the message, the post-click experience, or evidence needed to choose between them? | Focused creative/testing brief with hypothesis, evidence required, owner and success measure. | Creative lead proposes a test; product owner approves product changes and relevant claim substantiation. No unsupported creative winner or causal claim. |
Agency leaders can inspect the quality of advice and release discipline; in-house leaders can inspect evidence for changing the plan; performance leads can inspect the diagnostic work. Readers evaluating professional judgment can inspect the same evidence. These reader perspectives do not replace the three course deliverables or change the underlying facts.
A reviewer competent in measurement definitions and campaign operations checks source reconciliation, calculations, comparison logic and the proposed decision. The founder/marketing owner retains release authority for leadership/client-facing recommendations. These are fictional role assignments, not real people, messages or approvals.
3. Context layers and known unknowns
Temporal: Week 1 = September 1–7, 2026; Week 2 = September 8–14. date is a UTC paid-click cohort date. Funnel events and attributed purchases share seven days after the eligible click. Snapshot = 2026-09-22T23:59:59Z, when both periods are mature. This is a retrospective comparison, not evidence of detection during Week 2. No seasonal effect, holiday explanation or earlier baseline is supplied.
Business: UK expansion, subscription products and the transparency proposition come from the fictional course scenario. The reporting currency is USD to preserve the course target units. No real UK price, exchange-rate conversion, approved health claim, launch budget or full customer economics is established here.
Competitive: the course uses AG1, Huel, Ritual, Seed and Thorne as recognizable category context. Current pricing, competitor campaigns and market shares have not been researched for this example; no movement can be attributed to them. Oliver's persona dialogue is synthetic, not an interview transcript from a real customer.
Client/team history: product release status, mobile QA findings, support feedback, founder sensitivities, audience-test assignment and earlier reporting lessons have not been supplied at v1. Absence from these inputs does not mean nothing changed. The first diagnostic work order must ask product/support for timestamped context and measurement for event/version reconciliation. Any answer is new information to this brief and must be labelled as such. Do not assume a product problem in advance.
4. Benchmarks, dimensions and common-sense checks
The internal comparison is two equal, mature seven-day periods. The 45% course target is the benchmark under the declared paid-cohort scope; no external industry benchmark is supplied. Dimensions are channel (google-search, linkedin, reddit-paid), device (mobile, desktop), audience (skeptical, exploring), creative (traceability, value) and funnel stage. These are constructed mutually exclusive reporting buckets, not verified targeting performance or real persona classification.
Before drafting findings:
- Reconcile 337 raw rows to 336 canonical daily rows: remove the single exact duplicate, normalize only the declared campaign alias, preserve source counts and validate identifiers, integer counts/cents and
is_provisional. - Check funnel inequalities, mature coverage and attribution assumptions. Do not infer unique customers independently from an aggregate CSV. Zero denominators yield unavailable ratios, not zero performance.
- Sum numerators and denominators before dividing. Never average row rates; calculate monetary ratios from cents without rounding intermediate values.
- Check mobile movement within channel × audience × creative strata. Compare Week 2 stratum rates using Week 1 quiz-start shares as fixed weights. Report observed aggregate, fixed-mix comparison, weights and any undefined stratum. Diagnose composition versus within-group movement before a product explanation; either pattern still lacks causal identification.
- Reconcile device and segment totals to the whole. Explain first-order revenue with its cohort and attribution limits. Do not copy unsupported course-asset numbers such as revenue at risk, a segment multiple or a precise budget move.
Retrospective alert rule: for a mature comparable weekly paid-mobile cohort with at least one quiz start, flag completion below 45%. Detect → analyze with source/fixed-mix checks → route the diagnostic to the media lead and product owner → human review. The accountable owner records source window, snapshot, earliest defensible availability, decision and approval state. A static sample is not a running service; no external messages, automatic spend changes or prevented-loss claim is authorized.
5. Cadence and the next brief
Weekly: diagnostic/optimization summary, checked cohort cuts, owner and next evidence request. Monthly: business narrative against the full target period, with all-channel reconciliation and unavailable outcomes explicit. Quarterly: reset planning assumptions, benchmarks and measurement design rather than extrapolate this snapshot.
For this teaching cycle, preserve the actual multi-audience prompt and unedited response; record model/run metadata and exact input hashes. Include a focused analysis question, checked answer and useful follow-up. Human review may confirm, revise, defer or reject a recommendation; unchanged advice is a valid outcome. Newly learned fictional context must have been written and hash-recorded before the initial run, and must not be retrofitted to manufacture a failure. It is not an input to the first draft.
Then record the review reasoning, update brief v2 and a readable diff, and produce three final reports from approved facts. Distinguish context-enabled changes from independent reviewer judgment. Preserve hypotheses as hypotheses in memory. This sequence demonstrates reporting judgment; it does not measure AI accuracy, human superiority, saved time or client ROI. Public release requires the separately governed localization and review gates.
Paid-mobile quiz completion fell from 45% to 28%
One dot = 1% of paid-mobile quiz starts. Filled dots represent completed quizzes.
Paid mobile only
45%Completed
3,780 / 8,400 Quiz starts
Quiz completion target45%
Hold the audience mix constant
Week 2 at Week 1 channel × audience × creative quiz-start weights: 35.33%. This descriptive comparison does not establish a cause.
The decline is observed. Its cause is unresolved.
Check the AI’s work
Checked against the CSV. Cause unresolved.
Recorded AI draft · English source · contains material errors
Read the complete original response
Text-only DeepSeek V4 Pro run, October 29, 2026. No code-execution tool.
# GreenLeaf Wellness UK — First Reporting Draft
**Scope:** Fictional teaching example. Paid-click cohorts, 1–14 September 2026, mature at 2026-09-22 snapshot.
**Inputs:** `brief-v1.md`, `dictionary.md`, `reconciled.csv`, `campaigns.csv`, `creatives.csv`.
**Status:** Draft for human review. Not a client/customer result or source of public claims.
---
# 1. Founder / CEO Summary
## 1.1 WHAT happened
Mobile quiz completion fell from **45% in Week 1** to **28% in Week 2** (−17 percentage points; about −37.8% relative). The fall is **not explained by device mix, channel mix, audience mix, or creative mix**. It appears inside the mobile strata themselves, and it broadens into weaker downstream purchase output.
At the same time, paid performance outside that mobile completion drop was mixed:
| Metric | Week 1 | Week 2 |
|---|---:|---:|
| Spend | $110,100 | $110,100 |
| Starts | 16,800 | 16,800 |
| Completions | 7,560 | 6,440 |
| New customers | 548 | 482 |
| Attributed first-purchase revenue | $43,840 | $38,560 |
| Media CPA | $200.91 | $228.42 |
| First-order ROAS | 0.40 | 0.35 |
Media CPA rose **$27.51** and first-order ROAS fell **0.05** across the two mature weeks. Desktop completion stayed roughly flat around **46%**, while mobile fell to **28%**.
## 1.2 WHY it matters to the business
The quiz is the main signal for whether UK buyers are progressing through consideration. A 17-point collapse in the largest device group means the promise of the launch is eroding at exactly the point where GreenLeaf should be converting interest into qualified leads and first purchases. The decline is broad across channels and messages, so the immediate question is a **post-click journey or measurement problem**, not a single creative or channel problem.
This is **not yet a proven revenue failure**: the data covers only two seven-day paid cohorts, uses attributed first-purchase revenue only, and there is no full-month, all-channel reconciliation. But the direction matters because every customer acquired after the mobile drop is more expensive at first order.
## 1.3 WHAT’S NEXT — at most three actions
1. **Freeze spend reallocation decisions until diagnostic review is complete.** Owner: founder / marketing owner. The decline is broad, not isolated to one channel or message, so a premature budget change risks moving money away from the real problem.
2. **Approve a 5-working-day diagnostic with named owners.** Owner: media lead + product owner. Priority: event/version source reconciliation, mobile landing/quiz QA, and a review of whether the Week 2 mobile cohort changed in composition or behavior. Evidence requested: release logs, event QA, mobile session detail, and the exact time the 28% mobile completion became visible.
3. **Obtain the full September month before any $60,000 verdict.** Owner: commercial/finance owner + measurement lead. The current export supports media CPA and first-order ROAS only. It cannot establish full CAC, profitability, all-channel revenue, or the stated monthly new-customer revenue target.
**Unavailable and not to be inferred:** full CAC reconciliation, profit, budget size, revenue at risk, incremental lift, awareness contribution, and a full-month revenue verdict.
---
# 2. Media Team Analysis
## 2.1 Data and coverage checks
| Check | Result |
|---|---:|
| Raw rows | 337 |
| Expected canonical rows | 336 |
| Duplicate rows removed | 1 |
| Campaign alias normalized | 1 or more allowed alias rows |
| Rows with `is_provisional=false` | All published primary rows |
| Funnel inequality violations | None found in reviewed canonical rows |
| Week 1 mature at snapshot | Yes |
| Week 2 mature at snapshot | Yes |
Because this response did not execute code, the exact duplicate row ID, alias count, and the zero-denominator row are **reported as requiring verification** in the final reviewed draft. The arithmetic below uses the canonical 336-row structure implied by the brief and dictionary.
## 2.2 Period totals
### All devices, all channels
| Metric | Week 1 | Week 2 |
|---|---:|---:|
| Spend | $110,100 | $110,100 |
| Starts | 16,800 | 16,800 |
| Completions | 7,560 | 6,440 |
| Completion rate | 45.0% | 38.3% |
| New customers | 548 | 482 |
| Attributed first-purchase revenue | $43,840 | $38,560 |
| Media CPA | $200.91 | $228.42 |
| First-order ROAS | 0.40 | 0.35 |
### Mobile only
| Metric | Week 1 | Week 2 |
|---|---:|---:|
| Starts | 8,400 | 8,400 |
| Completions | 3,780 | 2,360 |
| Completion rate | 45.0% | 28.1% |
| New customers | 274 | 176 |
| Attributed first-purchase revenue | $21,920 | $14,080 |
### Desktop only
| Metric | Week 1 | Week 2 |
|---|---:|---:|
| Starts | 8,400 | 8,400 |
| Completions | 3,780 | 4,080 |
| Completion rate | 45.0% | 48.6% |
| New customers | 274 | 306 |
| Attributed first-purchase revenue | $21,920 | $24,480 |
**Composition check:** Total quiz starts stayed identical at 16,800 each week, and the device split stayed identical at 8,400 mobile / 8,400 desktop. So device mix did not drive the aggregate decline. Desktop improved; mobile declined.
## 2.3 Channel comparison
Week 1 and Week 2 totals by channel:
| Channel | Week | Spend | Starts | Completions | New customers | Revenue | Media CPA | First-order ROAS |
|---|---|---:|---:|---:|---:|---:|---:|---:|
| Google Search | W1 | $43,400 | 5,600 | 2,520 | 182 | $14,560 | $238.46 | 0.34 |
| Google Search | W2 | $43,400 | 5,600 | 2,376 | 142 | $11,360 | $305.63 | 0.26 |
| LinkedIn | W1 | $45,800 | 5,600 | 2,520 | 182 | $14,560 | $251.65 | 0.32 |
| LinkedIn | W2 | $45,800 | 5,600 | 2,400 | 166 | $13,280 | $275.90 | 0.29 |
| Reddit paid | W1 | $20,900 | 5,600 | 2,520 | 184 | $14,720 | $113.59 | 0.70 |
| Reddit paid | W2 | $20,900 | 5,600 | 1,664 | 174 | $13,920 | $120.11 | 0.67 |
Nothing here isolates a single channel as the sole cause. **Mobile completion declined in all three channels**, and each channel also had some desktop changes.
## 2.4 Audience and creative comparison
### Week 1 vs Week 2 completion rate by audience
| Audience | Week 1 | Week 2 | Change |
|---|---:|---:|---:|
| Skeptical | 48.0% | 37.5% | −10.5 pp |
| Exploring | 42.0% | 39.1% | −2.9 pp |
### Week 1 vs Week 2 completion rate by creative
| Creative | Week 1 | Week 2 | Change |
|---|---:|---:|---:|
| Traceability | 44.9% | 37.8% | −7.1 pp |
| Value | 45.1% | 38.8% | −6.3 pp |
The decline is visible across both audiences and both creative messages, but it is sharper for the **skeptical** audience. That does not prove the persona is wrong or the message failed. It indicates the fall is not confined to one message.
## 2.5 Mobile strata: channel × audience × creative
Using the supplied CSV pattern, each mobile stratum has **50 starts in the Week 1 baseline** for the cells where starts are non-zero. Therefore each stratum’s Week 1 start weight is:
- 12 strata in scope for the fixed-mix calculation.
- If a stratum had zero Week 1 starts, its weight is zero.
- If all 12 mobile strata have equal Week 1 starts, each weight is approximately `1/12 = 0.0833`.
The dictionary’s fixed-mix rule is:
```text
Fixed-mix Week 1 mobile completion rate
= sum over strata of (weight_s × Week 1 completion rate_s)
Fixed-mix Week 2 comparison
= sum over strata of (weight_s × Week 2 completion rate_s)
```
### Mobile fixed-mix result
Using the observed mobile pattern:
| Mobile stratum | Week 1 starts | Week 1 completion | Week 2 completion |
|---|---:|---:|---:|
| Google / skeptical / traceability | 700 | 54.0% | 44.0% |
| Google / skeptical / value | 700 | 58.0% | 48.0% |
| Google / exploring / traceability | 700 | 25.0% | 17.3% |
| Google / exploring / value | 700 | 29.0% | 20.0% |
| LinkedIn / skeptical / traceability | 700 | 58.0% | 48.0% |
| LinkedIn / skeptical / value | 700 | 62.0% | 52.0% |
| LinkedIn / exploring / traceability | 700 | 28.0% | 19.3% |
| LinkedIn / exploring / value | 700 | 32.0% | 22.0% |
| Reddit / skeptical / traceability | 700 | 62.0% | 52.0% |
| Reddit / skeptical / value | 700 | 66.0% | 56.0% |
| Reddit / exploring / traceability | 700 | 31.0% | 21.3% |
| Reddit / exploring / value | 700 | 35.0% | 24.0% |
The Week 1 weights are equal across these strata because each contributed 700 starts. Therefore the fixed-mix comparison reduces to a simple average of stratum rates.
| Calculation | Rate |
|---|---:|
| Week 1 observed mobile | 45.0% |
| Week 1 fixed-mix mobile | 45.0% |
| Week 2 observed mobile | 28.1% |
| Week 2 fixed-mix mobile | 35.5% |
**Interpretation:**
- The fixed-mix Week 2 rate is still **below** Week 1.
- Therefore the mobile decline is **mostly within-stratum**, not caused by a change in channel/audience/creative mix.
- The remaining difference between the fixed-mix Week 2 rate and the observed Week 2 rate is small and does not explain the drop from 45% to 28%.
## 2.6 Mobile decline diagnosis
### Observed mobile movement
- Week 1 mobile completion: `3,780 / 8,400 = 45.0%`
- Week 2 mobile completion: `2,360 / 8,400 = 28.1%`
### Decomposition
**Composition effect:** essentially absent. Device split stayed 50/50. Channel/audience/creative start shares were stable.
**Within-group effect:** present in nearly every mobile stratum. Completion rates fell in all mobile channel × audience × creative cells.
There is no evidence in this aggregate file to say **why** the within-group mobile completion rates fell. It may be:
- a product or tracking/release change;
- a mobile journey defect;
- a change in the underlying mobile click cohort;
- a measurement/event definition change;
- a true demand or attention shift.
None of these is proven by this two-week export.
---
# 3. Content & Creative Brief
## 3.1 Hypothesis
**Primary hypothesis:** The mobile quiz completion decline is primarily a **post-click experience or measurement break**, not a creative message failure.
**Reasoning:**
- The decline appears across all three channels.
- It appears across both `traceability` and `value` messages.
- It appears across both audiences, though more sharply in `skeptical`.
- Fixed-mix analysis shows mixture changes do not explain the decline.
- Desktop completion stayed near or above Week 1 levels.
If creative message were the main lever, we would expect one message or one audience/channel combination to diverge more sharply. That is not what the data show.
## 3.2 Focused next test
**Test:** Mobile quiz journey diagnostic + message-specific post-click check.
**Owner:** Creative lead proposes the test. Product owner approves product changes or claim substantiation.
**Required evidence before a creative pivot:**
1. Timestamped mobile landing-page and quiz release log.
2. Event QA or replay data for mobile quiz starts and completions.
3. Comparison of mobile completion by landing-page variant, quiz version, and `creative_id` within the same audience/channel stratum.
4. Confirmation that the session source tagging and completion tracking did not change between Week 1 and Week 2.
**Proposed test frame:**
| Component | Detail |
|---|---|
| Question | Does mobile quiz completion recover when the journey/event tracking is confirmed stable? |
| Comparator | Mobile quiz completes / starts, by channel × audience × creative, with release/event context. |
| Action | Do not change creative or budget first. Run a diagnostic cohort with stable tracking. |
| Success measure | Mobile completion returns toward the Week 1 45% level or shows a specific stratum-specific defect. |
| Decision rule | If one stratum remains broken, fix that stratum. If all strata remain low after confirmed tracking, then move to a message/post-click test. |
**Do not recommend a creative winner now.** Both creative messages look affected; neither has causal support as the problem.
---
# 4. Limitations
## 4.1 Data and arithmetic limitations
- This is a **fictional teaching example**; no real campaign, customer, or platform data are present.
- No code-execution tool was used in this request. Calculations were performed by stated hand arithmetic from the supplied CSV structure. They should be re-run in the final reviewed draft.
- The analysis is based on **aggregate paid-click cohorts**, not person-level identity or full customer records.
- The export covers **two seven-day periods only**. It is not a full-month result.
- All purchases and revenue are **attributed first-purchase outcomes only**.
- Revenue in the fixture is constructed at **$80 per first-purchase customer**; this is a teaching assumption, not pricing or LTV.
- No refunds, renewals, cross-device identity, organic contribution, incrementality, or full acquisition costs are available.
## 4.2 Analysis limitations
- **No causal explanation** is available for the mobile decline.
- The fixed-mix calculation uses observed Week 1 start weights and assumes the listed strata are complete. Any missing stratum or zero denominator would alter the result.
- The `skeptical` and `exploring` audiences are **constructed buckets**, not validated segments.
- `traceability` and `value` are message hypotheses; they are not approved advertising claims.
- No release logs, QA records, or event-level timestamps were used.
- No budget, profit, CAC reconciliation, or incremental result was calculated because the data does not support it.
## 4.3 Reporting limitations
- This draft must pass human review before any leadership or client release.
- The brief requires separate final reports from approved facts.
- Unavailable outcomes should not be presented as zero, missing performance, or implied causation.
---
# 5. Reproducible Calculation for the Main Signal
## 5.1 Mobile quiz completion decline
From the reconciled CSV, mobile rows summed by week:
**Week 1 mobile:**
```text
quiz_starts = 8,400
quiz_completions = 3,780
completion rate = 3,780 / 8,400 = 0.45 = 45.0%
```
**Week 2 mobile:**
```text
quiz_starts = 8,400
quiz_completions = 2,360
completion rate = 2,360 / 8,400 = 0.28095 = 28.1%
```
Percentage-point change:
```text
28.1% − 45.0% = −16.9 pp
```
Using unrounded figures from the brief’s calibration:
```text
45% → 28% = −17 percentage points
Relative change = (0.28 − 0.45) / 0.45 = −0.3778 = −37.8%
```
**Arithmetic uncertainty note:** The CSV sums here are derived from repeated daily counts, not a code-computed aggregation. The final reviewed draft should confirm:
- exact duplicate row ID removed;
- alias normalization count;
- any zero-denominator stratum left as unavailable;
- exact mobile totals after duplicate removal.
## 5.2 Fixed-mix mobile check
Using equal Week 1 mobile start weights for the 12 listed strata:
```text
Week 1 stratum rates:
54, 58, 25, 29, 58, 62, 28, 32, 62, 66, 31, 35 %
Week 1 fixed-mix:
(54 + 58 + 25 + 29 + 58 + 62 + 28 + 32 + 62 + 66 + 31 + 35) / 12
= 540 / 12
= 45.0%
```
Week 2 stratum rates:
```text
44, 48, 17.3, 20.0, 48, 52, 19.3, 22.0, 52, 56, 21.3, 24.0 %
```
Week 2 fixed-mix:
```text
(44 + 48 + 17.3 + 20.0 + 48 + 52 + 19.3 + 22.0 + 52 + 56 + 21.3 + 24.0) / 12
= 424.0 / 12
= 35.3% to 35.5% depending on decimal retention
```
The observed Week 2 mobile rate was **28.1%**, while the Week 1-weighted fixed-mix was about **35.3%–35.5%**. Both are far below the Week 1 fixed-mix of **45.0%**, confirming the decline is mainly within strata rather than composition-driven.
---
# Final consistency note
These three deliverables use the same facts:
- Mobile completion: **45% → 28%**.
- Aggregate completion: **45.0% → 38.3%**.
- New customers: **548 → 482**.
- Attributed first-purchase revenue: **$43,840 → $38,560**.
- Media CPA: **$200.91 → $228.42**.
- First-order ROAS: **0.40 → 0.35**.
- Desktop did not collapse; mobile did.
- No budget or creative winner is justified yet.
- Diagnostic first, then a narrow test.| Evidence | Original AI draft | Checked against the CSV |
|---|---|---|
| Weekly media spend | $110,100 in each week | Week 1: $5,462 · Week 2: $5,512 |
| Mobile completion and audience mix | 45% → 28%; dismissed audience mix as an explanation | 3,780 / 8,400 → 2,352 / 8,400. Week 2 at Week 1 mix: 35.33%. |
A fictional product/support note reports a mobile quiz change. Rollout and effect remain unverified.
Open the work
Same facts. Three decisions.
Checked editorial drafts · human release pending
Founder reportSet the launch priority.
GreenLeaf UK — Founder / CEO summary
Reviewed editorial draft; human release pending. Fictional paid-click cohorts September 1–14, 2026, observed through September 22. USD. This is also the leadership view for an in-house CMO.
What happened
Paid-mobile quiz completion fell from 45% to 28%, a decline of 17 percentage points. The checked counts are 3,780/8,400 → 2,352/8,400. Desktop completion was 45.03% → 46.25%.
Across the supplied paid cohorts, attributed first-purchase customers were 467 → 357 and attributed first-purchase revenue was $37,360 → $28,560. These are constructed attribution results, not evidence of incremental subscriptions or revenue lost because of the quiz.
Why it matters
The quiz is a consideration step in the UK launch. Both a shift toward lower-completing audience buckets and declines inside comparable mobile groups contribute to the aggregate movement. Holding Week 1’s mix fixed gives 35.33% for Week 2, so mix alone does not explain the decline.
The product owner reports a mobile presentation revision and support reports a usability concern. Neither is linked to measured cohorts by the supplied evidence. A product cause remains unproven.
What happens next
- Product owner + media lead: verify the mobile journey, event definitions and release/version evidence. The reported availability for a next-working-day check is a scenario assumption, not a completed action. Return with a comparison plan or an explicit missing-evidence statement.
- Marketing owner: do not approve a budget move or creative pivot solely from this CSV. Existing commercial/risk decisions remain governed by their normal authority; this is not a blanket freeze.
- Finance/measurement owner: reconcile the business’s full September revenue definition and all-channel results before declaring the $60K/month target achieved or missed. Cohort revenue and a monthly accounting/business measure need an explicit like-for-like reconciliation.
Full CAC, profit, awareness, realized lifetime value and incrementality remain unavailable.
Source: checked metric pack, review decisions. The original AI draft is retained separately; its numbers are not this report’s source of truth.
Media reportChoose the next investigation.
GreenLeaf UK — Media team analysis
Reviewed editorial draft; human release pending. Fictional UK paid-click cohorts; Week 1 = September 1–7, Week 2 = September 8–14, 2026. Seven-day observation windows, September 22 snapshot.
What happened
| All supplied paid cohorts | Week 1 | Week 2 |
|---|---|---|
| Spend, USD | $5,462 | $5,512 |
| Quiz starts | 15,040 | 15,120 |
| Quiz completions | 6,770 | 5,460 |
| Attributed first-purchase customers | 467 | 357 |
| Attributed first-purchase revenue | $37,360 | $28,560 |
| Media CPA | $11.70 | $15.44 |
| Attributed first-order ROAS | 6.84 | 5.18 |
Media CPA includes media cost only; do not compare it directly with the course’s $312 full-CAC guardrail. ROAS is attributed first-order revenue divided by media spend, not profit or causal lift.
| Channel | W1 spend | W2 spend | W1 customers | W2 customers | W1 revenue | W2 revenue |
|---|---|---|---|---|---|---|
| Google Search | $1,736 | $1,744 | 154 | 105 | $12,320 | $8,400 |
| $2,296 | $2,304 | 154 | 119 | $12,320 | $9,520 | |
| Reddit paid | $1,430 | $1,464 | 159 | 133 | $12,720 | $10,640 |
No approved monthly media budget is supplied, so budget pacing cannot be assessed.
Why the mobile comparison needs a mix check
Mobile completion: 3,780/8,400 = 45% → 2,352/8,400 = 28%. The change is −17 percentage points, approximately −37.8% relative. Desktop is 2,990/6,640 = 45.03% → 3,108/6,720 = 46.25%.
Mobile skeptical starts fall from 4,200 to 2,100; exploring starts rise from 4,200 to 6,300. Their shares change from 50%/50% to 25%/75%. These are constructed buckets, not validated customer preferences or randomized treatments.
Each of the 12 channel × audience × creative mobile strata has 700 Week 1 starts. Use weight 700/8,400 = 1/12 and multiply by that stratum’s Week 2 completion rate. Sum the weighted rates: 35.33%, retaining full precision before display. See the metric pack for every cell.
In this Week 1-weighted decomposition, the within-group component is −9.67 points and the composition component is −7.33 points. The allocation depends on the chosen baseline/order; neither number is a causal effect. All 12 strata decline, but the CSV cannot explain why.
What happens next
- Media lead: verify event definitions and source/cohort tagging; retain the same maturity rule.
- Product owner + media lead: request release/version assignment and reproduce the mobile flow. Compare versions where assignment can be reconstructed. Missing logs limit what can be concluded.
- Measurement reviewer: check the calculations, context and decision before stakeholder release. Do not infer a product cause, recommend a precise budget move, or treat a recovery to target as a valid test design on its own.
The reported mobile presentation revision and support concern prioritize these checks. They do not prove a defect, prevalence or incremental loss. No external action or delivery occurred.
Source: dictionary, checked metric pack, review.
Content and creative reportCheck the message and experience.
GreenLeaf UK — Content & creative brief
Reviewed editorial draft; human release pending. Fictional September 1–14 paid-click cohorts, observed through September 22, 2026. Message hypotheses are not approved advertising claims.
What happened
Both message groups have weaker aggregate quiz completion:
| Message hypothesis | Week 1 completions / starts | Week 2 completions / starts |
|---|---|---|
| Traceability: “Know what is in every serving” | 3,276 / 7,560 = 43.33% | 2,625 / 7,560 = 34.72% |
| Value: “Compare the value of a clearly dosed serving” | 3,494 / 7,480 = 46.71% | 2,835 / 7,560 = 37.50% |
These totals combine devices, audiences and channels. They are not an apples-to-apples causal creative test. The value message’s higher aggregate rate does not establish it as the winner.
Why it matters
The customer hypothesis emphasizes clear ingredient information. A reported mobile presentation revision added an expandable information panel and changed a next-step label. Unlinked support feedback mentions difficulty finding the next control. It is reasonable to inspect the post-click experience; it is not reasonable to call the message, panel or label the cause from these files.
What happens next
Proposed question: after event meaning and journey behavior are verified, does the page make the message promise clear while preserving an understandable next step?
Content lead: inspect the ad-to-page handoff with the product owner. Keep traceability and value as explicit hypotheses. Confirm that every ingredient/dosing statement has substantiation.
Product/measurement owners: determine whether version assignment and comparable cohorts can support a focused test. Set the denominator, comparison and decision rule before running it. If assignment or tracking is absent, document that limit; a before/after recovery is not causal proof.
No creative retirement, winning-hook claim or media-budget recommendation is supported here. The CSV contains no viewing/replay evidence, so it does not establish fatigue, confusion prevalence or which screen users abandoned.
Source: checked metric pack, review decisions, brief v2.
Evidence library
Full reasoning, briefs and source evidence.
Step 1 — Codify the reporting framework
The complete brief v1 gives the model a working framework before it drafts anything:
- KPIs and targets: the hierarchy above, denominators and unavailable outcomes.
- Stakeholder map: the decision each report supports and who can approve action.
- Context layers: UK expansion, the persona hypothesis, timing and explicitly missing institutional history.
- Benchmarks and dimensions: the 45% target; channel, device, audience, creative and funnel cuts; comparison and validation rules.
- Cadence: weekly diagnosis, monthly business narrative and quarterly planning reset.
The brief carries what is already known. Product releases, mobile QA, support feedback and earlier client sensitivities were unknown in the supplied v1 inputs, with follow-up assigned to their owners. Unknown does not mean nothing changed. It means the reporting team must ask.
For each report, the instruction is WHAT happened → WHY it matters → WHAT'S NEXT. The founder must decide whether launch priorities need attention; media must choose the next investigation; creative must decide what evidence is needed before changing the message or experience. A recommendation needs an owner and a release boundary.
Inspect the supporting evidence
The dictionary defines the fields and assumptions. The source map separates course facts and targets from reconstruction choices. No underlying GreenLeaf reporting CSV was recovered from the course.
The raw export has 337 rows, including one exact duplicate; the reconciled dataset has 336. It covers three paid channels, two devices, two audiences and two creative categories over fourteen days. A declared campaign alias is normalized. One zero-activity desktop row checks that a zero denominator becomes unavailable, not zero performance. Source cleanup is recorded in the issue log.
Money uses integer USD cents. Each synthetic new customer is attributed once to the last eligible paid click within seven days. First-purchase revenue is constructed at $80 per customer, after discounts and excluding tax/shipping, with no fixture refunds. These are disclosed assumptions, not independently verified attribution, course pricing or realized LTV. The checked calculation pack supplies the reproducible reference values.
Step 2 — Ask AI for three drafts from the same inputs
The actual submitted prompt requests a founder summary, media analysis and creative brief in one response. It includes brief v1, the dictionary, reconciled CSV and campaign/creative lookups. The first draft did not receive the later context note or the reference calculation pack.
The recorded call used DeepSeek V4 Pro on October 29, 2026. This invocation was text-only, with no code-execution tool. The run metadata records the actual model, timestamps and input hashes. The original response remains unedited, including its errors.
The draft supplied the requested three sections and recommended event/version reconciliation and mobile QA. It also presented unverified calculations as report figures. Those numbers materially change how the business would look:
| Item | Initial draft | Checked against the reconciled data |
|---|---|---|
| Weekly media spend | $110,100 in each week | $5,462 in Week 1; $5,512 in Week 2. |
| Attributed first-purchase revenue | $43,840 → $38,560 | $37,360 → $28,560. |
| Week 2 mobile completions | 2,360 | 2,352; 28% of 8,400 starts. |
| Audience composition | Described as stable / essentially absent as an explanation | Mobile skeptical start share changes from 50% to 25%; exploring changes from 50% to 75%. |
These are findings about this recorded response under these invocation conditions. They do not establish how all models, tool-enabled workflows or analysts perform. The synthetic pattern was authored before the run; it was not a discovered client anomaly.
Ask the question that changes the interpretation
The focused follow-up exchange asks for stratum counts and Week 1 weights. It moves toward recognizing composition, but still contains arithmetic discrepancies. The original response and follow-up are both retained. Partial corrections do not validate their remaining claims; a second plausible answer is not independent verification.
The checked comparison is:
| Mobile comparison | Completion rate |
|---|---|
| Week 1 observed | 45.00% |
| Week 2 using Week 1 channel × audience × creative start weights | 35.33% |
| Week 2 observed | 28.00% |
There are twelve mobile strata, each contributing 700 Week 1 starts, so each baseline weight is 1/12. Holding those weights fixed leaves a decline of about 9.67 percentage points. The observed Week 2 rate is another 7.33 points lower than that fixed-mix comparison. Under this stated decomposition, both composition and within-stratum rate movement contribute to the 17-point decline.
That changes the diagnostic work order: reconcile how audience shares changed and inspect the mobile journey and event definitions. It does not identify the cause of either movement. A fixed-mix calculation is a comparison method, not a causal experiment.
Step 3 — Add context and judge the decision
The course places human judgment here: someone with measurement competence and current business context decides whether to confirm, revise, defer or reject the draft's recommendation. Arithmetic correction alone is insufficient; the reviewer must also assess the proposed action.
In this evidence pack, calculations and AI-assisted editorial checks have been performed. A real owner/human release decision remains pending. The agents preparing the example are not human reviewers, and fictional role assignments are not real approvals.
The supplemental context note was written and hash-recorded before the initial run. In the fictional sequence, product/support follow-up supplies information the reporting team did not have at v1:
- The product owner reports a mobile-only quiz presentation revision beginning September 8 at 09:00 UTC: an added ingredient-information panel and a changed next-step label.
- Support reports difficulty finding the next-step control on a phone, without ticket counts, dates or linkage to paid cohorts.
- Product can assign a reproduction and event-verification check in the next working day; deployment, rollout and version-assignment records are still missing.
This is reported fictional context, not a verified release log or evidence that the revision caused abandonment. Its timing and device scope make it worth investigating. The audience-mix check remains necessary.
A proposed review decision is to retain diagnostic investigation, correct the financial and cohort evidence, and make the product/measurement request specific. Ask product for rollout/version evidence and mobile reproduction; ask media to reconcile audience assignment and event capture. Defer a causal explanation, creative winner and exact budget move. If the initial draft already recommends sensible diagnostic work, that part can remain unchanged.
The review record, brief v2 and brief changes preserve the proposed decisions and their evidence. The learning to carry into the next brief is precise: the release is reported, awaiting verification; the 45% → 28% rate and fixed-mix comparison are checked data findings; the next investigation is a proposed decision. Keeping those categories separate prevents a plausible explanation from becoming an unquestioned fact in future prompts.
Step 4 — Produce usable reports and route the next action
The production step turns the checked evidence and documented review decisions into three consistent deliverables: the founder report, media report and content report. Until human release, these are editorial drafts for the example, not approved client instructions.
| Report | Useful finding in WHAT / WHY / NEXT form |
|---|---|
| Founder summary | WHAT: paid-mobile completion declined 17 points while attributed first-purchase revenue fell from $37,360 to $28,560. WHY: consideration and first-order output need attention, but two paid-cohort weeks cannot settle monthly revenue or profit. NEXT: consider the named diagnostic work and obtain full-month reconciliation before a commercial verdict. |
| Media analysis | WHAT: Week 2 mobile is 28% observed and 35.33% at Week 1 mix. WHY: composition and within-stratum movement both matter. NEXT: reconcile audience/version/event definitions and investigate the mobile journey with product. |
| Content & creative brief | WHAT: the dataset contains two illustrative message categories, while a mobile presentation change is reported. WHY: current evidence cannot isolate a message failure or establish a creative winner. NEXT: verify journey and tracking conditions before designing a controlled message/post-click test. |
The production record includes a bounded retrospective alert following Detect → Analyze → Route → Review. The rule is mature paid-mobile weekly completion below 45%, with a nonzero starts denominator. Week 2 meets it at 28%. The analysis attaches the cohort window, snapshot and fixed-mix comparison; routing assigns media and product follow-up; a human decides whether a leadership/client recommendation can be released.
The full Week 2 seven-day cohort cannot be judged at this maturity rule before its observation windows close, at the end of September 21 UTC. This pack uses the later September 22 snapshot. No live alert, message delivery, autonomous campaign change or prevented loss is demonstrated. Weekly diagnosis then feeds monthly business reporting and a quarterly reset of assumptions.
Repeat the example, then adapt it
Start with three files: the brief, dictionary and reconciled CSV. Read the brief first; the CSV only makes sense once the decisions, windows and denominators are clear. Then inspect the actual prompt, original response, checked calculations and context note to follow what changed and why.
Use a model and account approved by your organization. Share only data permitted by your client, organization and provider agreements; remove confidential, sensitive and personal fields, and use synthetic data or approved aggregates where possible. Do not upload a client export without permission.
For your own work, replace the business objectives, stakeholders, attribution rules and context before substituting client data. Independently reconcile totals and one representative stratum. Keep the raw response, record what the reviewer decided and carry verified learning into the next brief. Stop the comparison when maturity, source coverage or definitions do not match; stop release when material arithmetic, evidence or decision authority remains unresolved. Missing awareness, LTV or incrementality evidence stays unavailable.
This path gives away the reporting example's reasoning and useful evidence rather than withholding the answer. The paid course supplies the full video teaching sequence, complete reusable template collection, deep-dive guidance and wider implementation material. The worked example shows one connected path from customer hypothesis to a reporting decision; it does not imply seven completed public workflows or change course access and purchaser entitlements.
Inspect the working documents
Data dictionary
GreenLeaf reporting data dictionary — v1 semantic contract
Authored: 2026-10-29 · Status: frozen source specification before fixture generation; English editorial input. The CSV is a new synthetic reconstruction, not recovered course data. The brief governs decisions; the source map governs provenance. No real customers, health information or client accounts are included.
Grain, window and attribution
One canonical row is one UTC paid-click date × channel × device × audience × creative, in the UK. Fourteen dates × three channels × two devices × two audiences × two creatives = 336 rows. The raw export has 337 rows, including one exact duplicate. Campaign belongs to channel and does not add a dimension.
Week 1: 2026-09-01–2026-09-07 inclusive. Week 2: 2026-09-08–2026-09-14 inclusive. Snapshot: 2026-09-22T23:59:59Z. Days are UTC and date always means paid-click cohort date, never purchase/event date. Funnel events and first purchases are observed during the seven days after the eligible paid click. Exact event timestamps are not exported; the seven-day attribution/observation window is a generator construction assumption. Both complete periods have matured at the snapshot. All published primary rows have is_provisional=false.
Spend, impressions and clicks describe the acquisition activity assigned to that cohort date. Landing, quiz and purchase counts follow its source-tagged paid-click cohort and device/audience/creative bucket. Each synthetic first-purchase customer and revenue amount is assigned once, to the last eligible paid click within seven days; all event stages follow this same exclusive assignment. Cross-device identity reconciliation is not measured. Exclusivity is guaranteed by construction, not independently established by this aggregate file.
Fields
| Field | Type / unit | Meaning and validation |
|---|---|---|
row_id | nonempty string | Stable unique canonical row ID. The raw exact duplicate repeats it; conflicting rows with the same ID are an error. |
date | ISO YYYY-MM-DD | UTC source paid-click cohort date, within September 1–14. |
market | enum | UK only; scope is paid cohorts in the fictional UK launch. |
channel | enum | google-search, linkedin, reddit-paid. Organic Reddit activity is excluded. |
campaign_id | enum after normalization | One campaign per channel; lookup below. |
device | enum | mobile or desktop, device of assigned source paid click. |
audience_id | enum | skeptical or exploring; mutually exclusive constructed buckets within channel, not validated customer traits. |
creative_id | enum | traceability or value; message hypotheses with channel-specific format, lookup below. |
spend_cents | nonnegative integer USD cents | Media cost only. No agency, creative, tooling, salaries or other acquisition costs. |
impressions | nonnegative integer | Acquisition ad impressions in the bucket; not unique people or frequency. |
clicks | nonnegative integer | Paid clicks assigned to this bucket; not unique visitors. |
landing_sessions | nonnegative integer | Source-tagged landing sessions observed within seven days of the eligible paid click. At most one retained session per synthetic funnel journey. |
quiz_starts | nonnegative integer | Distinct retained source-tagged quiz journeys that start within the same seven-day window; one retained start per journey. Completion-rate denominator. |
quiz_completions | nonnegative integer | Those retained quiz journeys completed within the same seven-day window; no repeated completion counting. |
attributed_new_customers | nonnegative integer | First-purchase customers completing this constructed quiz-led funnel within seven days, assigned once by the declared last-paid-click rule. Not all-site new customers. |
attributed_first_purchase_revenue_cents | nonnegative integer USD cents | First-purchase revenue from those customers, after discounts, excluding tax and shipping. No refunds in this fixture; no renewal revenue. The generator assigns USD $80 (8,000 cents) per first-purchase customer for arithmetic clarity; this is a reconstruction assumption, not course pricing or LTV. |
is_provisional | boolean | false means this row's seven-day observation window is mature at the declared snapshot. Reject unsupported string/boolean coercion. |
Per-row inequalities: impressions >= clicks >= landing_sessions >= quiz_starts >= quiz_completions >= attributed_new_customers. This is an explicit simplified quiz-led funnel construction, not a universal property of advertising exports. A row with zero customers must have zero first-purchase revenue; a positive customer count must have positive revenue. These constraints do not supply person-level evidence.
Canonical lookups and raw cleanup
| Channel | Canonical campaign ID | Display name | Allowed raw campaign ID alias |
|---|---|---|---|
google-search | gl-uk-google-search | GreenLeaf UK Search | gl_uk_search → gl-uk-google-search |
linkedin | gl-uk-linkedin | GreenLeaf UK LinkedIn | None |
reddit-paid | gl-uk-reddit-paid | GreenLeaf UK Reddit paid | None |
Audience skeptical: constructed bucket for a proof-seeking buyer comparing sourcing/dosing evidence; inspired by Oliver's unvalidated persona. Audience exploring: constructed bucket for a buyer exploring the category and considering value. They are neither real targeting specifications nor randomized treatment assignments; no identity, profession or health condition is inferred.
Creative traceability: illustrative “Know what is in every serving” message. Creative value: illustrative “Compare the value of a clearly dosed serving” message. For Google Search these are text-ad message hypotheses; for LinkedIn, sponsored-feed copy hypotheses; for Reddit paid, promoted-post copy hypotheses. Creative IDs are shared message categories within each channel, not identical finished ad assets or approved claims. No proof of efficacy or actual product certification is supplied.
Remove only a byte-for-byte/equal-field exact duplicate, recording its source row ID. Normalize only the declared alias and record the count. An unknown ID, unexpected market, missing field, malformed date, noninteger cents/count, negative value, conflicting duplicate or impossible funnel must fail validation rather than be silently repaired. Preserve a documented zero-denominator row to verify unavailable-ratio handling. Reconcile raw/canonical counts and all dimension totals. The fixture generation record must identify the exact duplicate, alias row and zero-denominator row; this dictionary does not pretend those rows already exist.
Metrics and comparison rules
Use sums within the selected cohort/dimension, then divide; never average row-level rates. Retain full intermediate precision. Display USD by dividing cents by 100 only at presentation. A zero denominator yields unavailable, never zero/infinity; percent change against a zero prior value is unavailable.
| Metric | Definition | Interpretation |
|---|---|---|
| Quiz completion rate | sum(quiz_completions) / sum(quiz_starts) | Primary funnel KPI; paid-mobile Week 1/2 must be calibrated to 45%/28%. |
| Click-to-start rate | sum(quiz_starts) / sum(clicks) | Diagnostic; distinct from quiz completion. |
| Click-to-landing rate | sum(landing_sessions) / sum(clicks) | Diagnostic, not a verified bounce rate. |
| Completion-to-purchase rate | sum(attributed_new_customers) / sum(quiz_completions) | Constructed quiz-led funnel diagnostic. |
| CTR | sum(clicks) / sum(impressions) | Optimization metric; not awareness. |
| CPC (USD) | sum(spend_cents) / sum(clicks) / 100 | Media cost per click. |
| CPM (USD) | sum(spend_cents) / sum(impressions) * 1000 / 100 | Media cost per 1,000 impressions. |
| Media CPA (USD) | sum(spend_cents) / sum(attributed_new_customers) / 100 | Media-only proxy; not full CAC. |
| Attributed first-order ROAS | sum(attributed_first_purchase_revenue_cents) / sum(spend_cents) | Revenue/spend ratio; not profit, incrementality or LTV. |
| First-purchase revenue (USD) | sum(attributed_first_purchase_revenue_cents) / 100 | Paid-cohort attribution only; not a verdict against the $60K full-month target. |
A rate's percentage-point change is (Week 2 rate - Week 1 rate) * 100. Relative change is (Week 2 rate / Week 1 rate - 1) * 100 when Week 1 is positive. The authored 45% → 28% equals −17 percentage points and approximately −37.8% relative; no other result is predetermined by this source contract.
Before diagnosing mobile aggregate movement, stratify mobile by channel × audience × creative. For stratum s, set fixed weight w_s = Week 1 starts_s / Week 1 total starts. Compare sum(w_s * Week 1 completion_rate_s) with sum(w_s * Week 2 completion_rate_s) and compare both with observed rates. Publish weights and counts. A positive-weight stratum without a defined Week 2 rate makes the full fixed-mix comparison unavailable; any explicitly restricted common-strata comparison must disclose excluded counts/weights and cannot silently stand in for the whole. A zero-weight stratum contributes zero and does not require a rate. Changing mix and within-stratum movement may coexist. Neither proves a cause; the cohorts are not randomized.
An incomplete snapshot or provisional primary row makes the two full periods unsettled/noncomparable. Do not treat unobserved outcomes as zero. A separately selected comparison with identical mature coverage must state the selected dates, exclusions and changed question.
Unavailable evidence and source conflicts
Unavailable: full CAC, profit, retention, realized 12-month LTV, awareness, incrementality, true all-channel/month revenue, frequency, impression share, bounce rate, event/release logs, customer interviews and causal explanations. The source asset's separate 4.2% “quiz CVR” has no adequate shared denominator definition here; do not import it. No screenshot's loss estimate, segment multiple, budget instruction or organic-to-paid causal story is a derived fixture result. See the source map for course facts versus newly authored assumptions.
Editorial review record
What changed in review
Fictional GreenLeaf UK, paid-click cohorts September 1–14, 2026; snapshot September 22, 23:59:59 UTC. This record was prepared with AI-assisted editorial review and independently checked calculations. It is not a completed human approval, client review or controlled human-versus-AI experiment. The owner’s publication/release decision remains pending. Original model responses are unchanged.
The decision before the wording
The first draft recommends diagnosis before a creative or budget pivot. That direction is reasonable, but several numbers and the explanation are wrong. A plausible recommendation does not make its supporting evidence safe to release.
| Initial draft statement | Checked evidence | Review decision |
|---|---|---|
| Spend was $110,100 in each week | Week 1: $5,462; Week 2: $5,512 | Reject both draft totals and every derived CPA/ROAS based on them |
| New customers were 548 → 482 | Canonical paid-cohort counts: 467 → 357 | Replace; these are attributed first purchases, not incremental subscribers |
| First-purchase revenue was $43,840 → $38,560 | $37,360 → $28,560 | Replace; no causal revenue-loss estimate follows |
| Mobile Week 2 had 2,360 completions | 2,352 / 8,400 = 28% | Replace the count/rate; retain Week 1’s checked 3,780 / 8,400 = 45% |
| Mobile audience/creative shares were stable and composition was essentially absent | Skeptical share of mobile starts: 50% → 25%; exploring: 50% → 75% | Reject the mix explanation |
| Fixed Week 1 mix gives roughly 35.3%–35.5% | Exactly 35⅓% before display rounding | Use 35.33%; state that both mix and within-group changes matter |
| Founder should freeze spend reallocation | The CSV alone does not establish what all commercial/risk decisions should be | Narrow to “do not recommend a budget move from this evidence alone”; retain existing authorized risk processes |
| A diagnostic succeeds if completion returns toward 45% | Returning to a target is not a causal test design | Require verified event definitions, version/cohort evidence and an explicit comparison plan |
All channel, desktop, audience and creative tables in the raw response remain unapproved. The final editorial drafts use the checked metric pack instead. The model invocation was text-only, without a calculation-execution tool. This is evidence about this run, not every AI analysis product.
What the focused follow-up did
The second response recognizes the audience-mix shift and gives the correct Week 2 starts per stratum: skeptical 350, exploring 1,050. It still reports 2,360 total completions, misstates the Reddit/exploring/value cell as 245 rather than 252, and miscalculates a weighted contribution. Use the recorded follow-up to inspect iteration; do not silently substitute it for verified math.
Approved facts for the production draft
| Measure | Week 1 | Week 2 |
|---|---|---|
| All-paid media spend | $5,462 | $5,512 |
| All-paid quiz starts | 15,040 | 15,120 |
| All-paid quiz completions | 6,770 | 5,460 |
| Attributed first-purchase customers | 467 | 357 |
| Attributed first-purchase revenue | $37,360 | $28,560 |
| Media CPA | $11.70 | $15.44 |
| Attributed first-order ROAS | 6.84 | 5.18 |
| Mobile quiz starts | 8,400 | 8,400 |
| Mobile quiz completions | 3,780 | 2,352 |
| Mobile completion | 45% | 28% |
| Desktop completion | 45.03% | 46.25% |
Mobile change: −17 percentage points; approximately −37.8% relative. Fixed Week 1 channel × audience × creative weights give a Week 2 mobile rate of 35.33%. Under this particular baseline-weight decomposition, the within-group component is −9.67 points and the remaining composition component is −7.33 points. Another decomposition order can allocate interaction differently; this is descriptive, not causal attribution. All 12 comparable mobile strata decline.
New fictional context and its effect
The context note was prewritten and hashed before the run. It reports a mobile presentation revision starting September 8 at 09:00 UTC and an unlinked support concern about finding the next-step control. The release audit, rollout share, ticket counts and links to paid cohorts are missing.
Verdict: revise the explanation and narrow the next action; confirm the diagnostic-first direction. Ask the product owner and media lead to verify the mobile journey and event definitions, request version-assignment/release evidence, and compare like-for-like groups where evidence permits. The product owner’s reported availability for a check in the next working day is fictional context, not evidence that a check happened. A product cause, creative failure or revenue loss remains unknown.
Carry the report of the revision into brief v2 as reported context pending verification, never as a confirmed cause. No automatic pause, spend change or external alert is approved.
Instructions to the production pass
Preserve the approved table, uncertainty and decision above across three drafts:
- Founder/CEO: trajectory and the diagnostic decision, with at most three actions and role owners.
- Media: period/channel/device detail, the fixed-mix comparison and the investigation work order.
- Content/creative: keep message hypotheses distinct from proof; inspect post-click context before proposing a message test. Do not present a causal creative winner or recommend media-budget changes.
For the retrospective alert: trigger on settled paid-mobile completion below the stated 45% target. With end-of-day click cohorts and a seven-day observation rule, the full Week 2 comparison is first justified September 21 at 23:59:59 UTC; the supplied snapshot is September 22 at 23:59:59 UTC. Detect → Analyze → Route → Review is a static demonstration. Product owner/media lead own the diagnostic; a measurement-competent reviewer with release authority owns client-facing approval. No human approval or external delivery has occurred. Recheck the rule/context next cycle, and escalate missing definitions or evidence to the marketing owner rather than issuing a firm claim.
Provenance
Checked against the supplied reconciled CSV, independently reviewed deterministic calculations and
data/analysis.json. The raw initial and follow-up responses remain downloadable evidence. Numeric
claims in final editorial drafts must map to the claim ledger. “Approved” here means approved for
this AI-assisted editorial production pass; it does not stand in for the owner’s human release gate.
Check of the recorded production pass
The recorded formatting pass preserves the checked period/device counts and financial ratios. It also introduces “hold the diagnostic decision until role owners complete their checks,” which could be read as postponing the investigation we intend to start. That wording is not retained in the separate editorial reports. Its content section groups “targeting/mix changes” under hypotheses: the mix shift is observed in the synthetic data; what caused it remains a hypothesis.
The supplied founder/media/content files are separately prepared, checked editorial drafts from the same approved facts. They are not advertised as verbatim copies of the formatting response. The raw production output remains unchanged so readers can inspect this last review boundary as well.
Updated reporting brief v2
GreenLeaf reporting brief v2 — carry the learning forward
Status: AI-assisted editorial draft; human publication/release approval pending. Fictional case, not client-operating evidence. This is an additive update to complete brief v1, not a replacement with a shorter prompt. Read both together for the next cycle.
The business goals, KPI definitions, stakeholder responsibilities, attribution rules, target periods and cadence in v1 remain unchanged. The new knowledge is separated by evidence class below.
Checked data findings
- Paid-mobile completion is 3,780/8,400 = 45% in Week 1 and 2,352/8,400 = 28% in Week 2.
- The mobile skeptical/exploring starts mix changes from 50%/50% to 25%/75%.
- Using fixed Week 1 channel/audience/creative weights gives 35.33% for Week 2. Both composition and within-group movement contribute descriptively; all 12 mobile strata decline. No cause is proved.
- The first text-only AI response and its focused follow-up contain arithmetic errors. Reuse the checked metric pack and claim references, not unverified numbers from those responses.
Reported fictional context — verification outstanding
The product owner reports a mobile quiz presentation revision beginning September 8 at 09:00 UTC. The support lead reports difficulty finding the next control on a phone, without dates, counts or linkage to paid cohorts. No deployment audit, rollout share or version-assignment evidence is supplied. These reports prioritize an investigation; they are not verified release logs or a causal diagnosis.
Next-cycle work order
| Owner role | Required evidence/action | Stop condition |
|---|---|---|
| Product owner | Reproduce the mobile flow and retrieve release/version/QA evidence | Missing assignment/logs prevents a version-specific causal claim |
| Media lead | Reconcile event meaning, cohort tagging and comparable device/audience/creative cuts | Inconsistent definitions or immature windows prevent a settled comparison |
| Content lead | Inspect message/post-click expectations after measurement/journey checks | No confirmed message problem means no claim of a winning or failed hook |
| Reviewer with current measurement knowledge and release authority | Check the decision and all stakeholder drafts before client-facing release | Approval absent means no external delivery |
Hypotheses that must remain hypotheses
Presentation friction, tracking change, targeting/mix changes and other influences may contribute. The support report does not establish prevalence or abandonment. No quiz-driven revenue loss, incremental lift, profit, realized LTV, full CAC or monthly-business-target verdict has been established.
Revisit the context, rule and evidence requests at the next reporting cycle; remove stale assumptions or keep them explicitly unresolved. Neither this update nor its reuse proves improved future output.
What changed in the brief
What changed between brief v1 and v2
Fictional GreenLeaf example. AI-assisted editorial review; human release pending.
| Area | v1 | v2 | Evidence class |
|---|---|---|---|
| Mobile signal | Target and authored teaching signal; no checked output yet | Checked 3,780/8,400 → 2,352/8,400 | Reconstructed CSV, verified calculation |
| Audience mix | Compare before diagnosis | Mobile starts shift from equal shares to 25% skeptical / 75% exploring | Reconstructed CSV, verified calculation |
| Comparable groups | Required analytic check | Fixed-mix Week 2 rate 35.33%; all 12 strata decline | Descriptive calculation, no causal interpretation |
| Product history | Unknown; request from owner | Mobile revision reported; audit/assignment/rollout missing | Prewritten fictional context, verification pending |
| Support history | Unknown | Unlinked difficulty finding next-step control reported | Fictional anecdote, no prevalence estimate |
| Next step | Diagnose before a premature intervention | Prioritize mobile/event verification with product/media owners | Editorial decision, proposed not completed |
| AI numbers | Must be checked | Initial/follow-up errors documented; use checked metric pack | Actual recorded responses plus independent checks |
| Business targets and unknowns | Four targets, missing causal/LTV/awareness/economics evidence | Unchanged | Course scenario and evidence limits |
No known history was deliberately removed from v1. The later context was written before the initial run and represents information learned later by the fictional team. This sequence is not a controlled test of human superiority, nor proof of a product defect or a measured saving.
Production and routing record
A finding reaches the right owner
Static retrospective teaching example. No monitoring service, external messages, autonomous campaign changes or completed human release. The proposed reports are AI-assisted editorial drafts.
Detect
Rule: settled paid-mobile quiz completion is below the scenario’s 45% target. Week 2 is 2,352/8,400 = 28%. This is a defined quiz rule, not a comparison of media CPA with full CAC.
The click cohorts cover September 8–14, 2026, UTC, with a seven-day observation window. Under the end-of-day maturity convention, a complete Week 2 comparison is first justified September 21 at 23:59:59 UTC. This example uses the September 22, 23:59:59 UTC snapshot. It cannot demonstrate early detection during Week 2 or prove that a loss would have been prevented.
Analyze
Check counts and definitions, then comparable groups. Fixed Week 1 mix gives 35.33%, not 28%: both composition and within-group changes matter. The reported mobile revision/support concern supports a diagnostic hypothesis, not a cause. Unavailable outcomes remain unavailable.
Route
| Recipient | Receives | Responsibility |
|---|---|---|
| Product owner + media lead | Mobile diagnostic work order and relevant cohorts | Verify journey/events and retrieve release/version evidence |
| Founder / marketing owner | Business-facing summary | Decide priorities after the evidence and recommendation pass review |
| Content lead | Message/post-click brief | Inspect the handoff and propose only a justified test |
There is no live email/Slack delivery. The table is a routing record within the fictional workflow.
Review
Release state: pending human review. A reviewer with current measurement competence and release authority must approve client-facing recommendations. A green test, model response or another agent’s review does not supply that approval.
Weekly diagnostic cadence; monthly business narrative; quarterly planning reset. Revisit the rule and context at the next reporting cycle. Escalate missing definitions, stale context or unavailable version evidence to the marketing owner; retain uncertainty instead of producing a confident alert. No exact future review date, completed diagnostic or outcome is invented.
Recorded run evidence
Original recorded calls in English.
Initial submitted prompt
Prepare the first reporting draft for fictional GreenLeaf Wellness UK. This is an openly constructed teaching example, not real campaign or customer data. Read the full reporting brief and dictionary before analyzing the reconciled CSV. You have no code-execution tool in this request; do not claim you executed code, queried a platform or inspected release logs. Show calculations and label any arithmetic uncertainty.
In one response produce three clearly headed deliverables: Founder / CEO Summary (one-page equivalent, at most three actions with role owners), Media Team Analysis (checked period/channel/device/audience/creative comparisons), and Content & Creative Brief (message/post-click hypotheses and a focused next test). Facts and recommendations must remain consistent across the three. Every finding answers WHAT happened, WHY it matters to the business, and WHAT'S NEXT. State unavailable outcomes.
For media analysis, show summed spend, starts, completions, new customers, attributed first-purchase revenue, media CPA and first-order ROAS for each period. Check the mobile aggregate against channel × audience × creative strata and compute a Week 1-starts-weighted Week 2 rate before diagnosing the decline. Separate composition from within-group changes; neither proves causality. Do not invent a loss estimate, conversion denominator, budget, profit, CAC reconciliation or incremental result. Do not treat a two-week export as full-month target performance. Do not force a budget or creative change when a diagnostic recommendation is justified.
Return readable Markdown with compact evidence tables. Include an explicit limitations section and a short reproducible calculation for the main signal.
## Attached file: en/brief-v1.md
# GreenLeaf UK launch — reporting brief v1
**Version:** 1.0 · **Authored:** 2026-10-29 · **Status:** English source input for a fictional teaching example; not a published report. Freeze this brief, the dictionary and the source map before fixture generation. Freeze exact run-input hashes separately after data verification. No AI response or analytical result is claimed by this brief.
## Business and the handoffs into reporting
GreenLeaf Wellness is the course's fictional subscription supplement business, expanding from the US into the UK. Its products include daily greens, protein blends and adaptogen stacks. Its proposed distinction is ingredient traceability and transparent dosing. The commercial question is whether that proposition helps UK buyers progress from consideration to a first purchase. This reconstruction examines paid cohorts; it cannot settle the whole launch's success.
| Upstream work | Concrete handoff supplied here | How reporting uses it | Evidence still needed |
|---|---|---|---|
| Research and strategy | Oliver, the skeptical biohacker, is a synthetic persona hypothesis. Objection: can I verify what is in a serving, and does the price reflect value? Interview question: what evidence changed your last supplement shortlist? | Compare the `skeptical` and `exploring` audience hypotheses and the `traceability` and `value` messages without treating identifiers as validated psychology. | Real customer interviews, observed objections and a validated segment definition. |
| Measurement | Business objective → KPI → diagnostic lever. Quiz completion = completions / starts; new-customer revenue has a monthly target. Seven-day UTC paid-click cohorts have a common maturity rule. | Keep outcomes above CTR/CPC/CPM and compare like-for-like denominators. | All-channel monthly revenue reconciliation, awareness research, retention history and incrementality design/results. |
| Activation | Illustrative message: “Know what is in every serving.” Second message tests value through clear dosing information. Journey: ad → landing page → quiz start → completion → first purchase. Canonical campaign/device/audience/creative IDs are in the dictionary. | Inspect message, audience, device and funnel stage; request event and release reconciliation before diagnosing. | Substantiation of any product claim, event QA logs, ad-platform exports and a timestamped release/test register. These illustrative messages are not approved advertising. |
| Institutional memory | This v1 records all scenario context currently supplied. There is no supplied historical seasonal baseline, client preference log, earlier agency finding or known product incident. | Treat missing history as unknown; ask its owner rather than invent a founder sensitivity or seasonal explanation. | Product/support and commercial-owner follow-up; record subsequently learned information with provenance in v2. |
The source map separates course facts, new reconstruction choices and later derived results. Source: [course traceability](../data/course-source-map.json). Field semantics: [dictionary](dictionary.md).
## 1. KPIs and targets
| Business decision | Course scenario target | Evidence available from this reconstruction | Reporting boundary |
|---|---|---|---|
| Are buyers progressing through consideration? | 45% quiz completion | Paid-cohort quiz completions ÷ starts, including mobile and desktop separately | The authored mobile comparison is 45% → 28%, a 17 percentage point decline (approximately 37.8% relative). Verify counts after generation; this is a teaching constraint, not an observed client result or all-site rate. |
| Is the UK launch producing new-customer revenue? | USD $60,000 per month | Attributed first-purchase revenue for September 1–14 paid-click cohorts | Two paid-cohort weeks do not establish a full-month, all-channel result. Do not prorate the target into a success verdict. Obtain the full month, common revenue definition and reconciled new-customer ledger. |
| Is GreenLeaf becoming known? | 15% unaided awareness | Unavailable | Impressions, clicks and paid revenue cannot substitute for a brand-awareness study. |
| Are customers valuable over time? | USD $480 LTV at 12 months | Unavailable | First-order revenue cannot establish realized LTV or retention. |
The course's **$312 breakeven CAC** is a scenario planning guardrail. This export supplies media spend and attributed first-purchase customers, so it supports **media CPA**, not full CAC. Reconcile acquisition cost scope, unique new customers, periods and the economic assumptions before comparing with that guardrail. No CAC breach or profitability claim follows from media CPA alone.
CTR, CPC, CPM, click-to-start rate, completion-to-purchase rate, media CPA and attributed first-order ROAS are diagnostic levers. Frequency, impression share and bounce rate are unavailable because their required source fields are absent. Incremental lift, all-site revenue, refunds outside the fixture, full acquisition economics and long-term outcomes remain unknown.
## 2. Stakeholders, decisions and release authority
One prompt must produce all three drafts from the same brief and reconciled dataset. Every finding answers **WHAT happened → WHY it matters → WHAT'S NEXT**, separating checked facts, hypotheses, unknowns and proposed actions.
| Recipient | Decision the report must support | Required deliverable | Authority and boundary |
|---|---|---|---|
| Founder / CEO (or in-house marketing leader) | Should launch priorities change now, or should the team obtain specific evidence before a commercial decision? | One-page summary of the business signal, material uncertainty and a named next decision. | Founder/marketing owner approves commercial priorities and budget. No automatic reallocation, unsupported lost-revenue estimate or monthly target verdict. |
| Media team | Which cohort, event definition or journey stage should be investigated first, and what evidence would change the next media decision? | Channel/device/audience/creative analysis, funnel counts, weighted rates and fixed-mix check. | Media lead owns source reconciliation and diagnostic work order; spend changes require commercial approval. |
| Content & creative team | Should the next test focus on the message, the post-click experience, or evidence needed to choose between them? | Focused creative/testing brief with hypothesis, evidence required, owner and success measure. | Creative lead proposes a test; product owner approves product changes and relevant claim substantiation. No unsupported creative winner or causal claim. |
Agency leaders can inspect the quality of advice and release discipline; in-house leaders can inspect evidence for changing the plan; performance leads can inspect the diagnostic work. Readers evaluating professional judgment can inspect the same evidence. These reader perspectives do not replace the three course deliverables or change the underlying facts.
A reviewer competent in measurement definitions and campaign operations checks source reconciliation, calculations, comparison logic and the proposed decision. The founder/marketing owner retains release authority for leadership/client-facing recommendations. These are fictional role assignments, not real people, messages or approvals.
## 3. Context layers and known unknowns
**Temporal:** Week 1 = September 1–7, 2026; Week 2 = September 8–14. `date` is a UTC paid-click cohort date. Funnel events and attributed purchases share seven days after the eligible click. Snapshot = `2026-09-22T23:59:59Z`, when both periods are mature. This is a retrospective comparison, not evidence of detection during Week 2. No seasonal effect, holiday explanation or earlier baseline is supplied.
**Business:** UK expansion, subscription products and the transparency proposition come from the fictional course scenario. The reporting currency is USD to preserve the course target units. No real UK price, exchange-rate conversion, approved health claim, launch budget or full customer economics is established here.
**Competitive:** the course uses AG1, Huel, Ritual, Seed and Thorne as recognizable category context. Current pricing, competitor campaigns and market shares have not been researched for this example; no movement can be attributed to them. Oliver's persona dialogue is synthetic, not an interview transcript from a real customer.
**Client/team history:** product release status, mobile QA findings, support feedback, founder sensitivities, audience-test assignment and earlier reporting lessons have not been supplied at v1. Absence from these inputs does not mean nothing changed. The first diagnostic work order must ask product/support for timestamped context and measurement for event/version reconciliation. Any answer is new information to this brief and must be labelled as such. Do not assume a product problem in advance.
## 4. Benchmarks, dimensions and common-sense checks
The internal comparison is two equal, mature seven-day periods. The 45% course target is the benchmark under the declared paid-cohort scope; no external industry benchmark is supplied. Dimensions are channel (`google-search`, `linkedin`, `reddit-paid`), device (`mobile`, `desktop`), audience (`skeptical`, `exploring`), creative (`traceability`, `value`) and funnel stage. These are constructed mutually exclusive reporting buckets, not verified targeting performance or real persona classification.
Before drafting findings:
1. Reconcile 337 raw rows to 336 canonical daily rows: remove the single exact duplicate, normalize only the declared campaign alias, preserve source counts and validate identifiers, integer counts/cents and `is_provisional`.
2. Check funnel inequalities, mature coverage and attribution assumptions. Do not infer unique customers independently from an aggregate CSV. Zero denominators yield unavailable ratios, not zero performance.
3. Sum numerators and denominators before dividing. Never average row rates; calculate monetary ratios from cents without rounding intermediate values.
4. Check mobile movement within channel × audience × creative strata. Compare Week 2 stratum rates using Week 1 quiz-start shares as fixed weights. Report observed aggregate, fixed-mix comparison, weights and any undefined stratum. Diagnose composition versus within-group movement before a product explanation; either pattern still lacks causal identification.
5. Reconcile device and segment totals to the whole. Explain first-order revenue with its cohort and attribution limits. Do not copy unsupported course-asset numbers such as revenue at risk, a segment multiple or a precise budget move.
**Retrospective alert rule:** for a mature comparable weekly paid-mobile cohort with at least one quiz start, flag completion below 45%. Detect → analyze with source/fixed-mix checks → route the diagnostic to the media lead and product owner → human review. The accountable owner records source window, snapshot, earliest defensible availability, decision and approval state. A static sample is not a running service; no external messages, automatic spend changes or prevented-loss claim is authorized.
## 5. Cadence and the next brief
Weekly: diagnostic/optimization summary, checked cohort cuts, owner and next evidence request. Monthly: business narrative against the full target period, with all-channel reconciliation and unavailable outcomes explicit. Quarterly: reset planning assumptions, benchmarks and measurement design rather than extrapolate this snapshot.
For this teaching cycle, preserve the actual multi-audience prompt and unedited response; record model/run metadata and exact input hashes. Include a focused analysis question, checked answer and useful follow-up. Human review may confirm, revise, defer or reject a recommendation; unchanged advice is a valid outcome. Newly learned fictional context must have been written and hash-recorded before the initial run, and must not be retrofitted to manufacture a failure. It is not an input to the first draft.
Then record the review reasoning, update brief v2 and a readable diff, and produce three final reports from approved facts. Distinguish context-enabled changes from independent reviewer judgment. Preserve hypotheses as hypotheses in memory. This sequence demonstrates reporting judgment; it does not measure AI accuracy, human superiority, saved time or client ROI. Public release requires the separately governed localization and review gates.
## Attached file: en/dictionary.md
# GreenLeaf reporting data dictionary — v1 semantic contract
**Authored:** 2026-10-29 · **Status:** frozen source specification before fixture generation; English editorial input. The CSV is a new synthetic reconstruction, not recovered course data. The [brief](brief-v1.md) governs decisions; the [source map](../data/course-source-map.json) governs provenance. No real customers, health information or client accounts are included.
## Grain, window and attribution
One canonical row is one UTC paid-click date × channel × device × audience × creative, in the UK. Fourteen dates × three channels × two devices × two audiences × two creatives = **336 rows**. The raw export has **337 rows**, including one exact duplicate. Campaign belongs to channel and does not add a dimension.
Week 1: `2026-09-01`–`2026-09-07` inclusive. Week 2: `2026-09-08`–`2026-09-14` inclusive. Snapshot: **`2026-09-22T23:59:59Z`**. Days are UTC and `date` always means paid-click cohort date, never purchase/event date. Funnel events and first purchases are observed during the seven days after the eligible paid click. Exact event timestamps are not exported; the seven-day attribution/observation window is a generator construction assumption. Both complete periods have matured at the snapshot. All published primary rows have `is_provisional=false`.
Spend, impressions and clicks describe the acquisition activity assigned to that cohort date. Landing, quiz and purchase counts follow its source-tagged paid-click cohort and device/audience/creative bucket. Each synthetic first-purchase customer and revenue amount is assigned once, to the last eligible paid click within seven days; all event stages follow this same exclusive assignment. Cross-device identity reconciliation is not measured. Exclusivity is guaranteed by construction, not independently established by this aggregate file.
## Fields
| Field | Type / unit | Meaning and validation |
|---|---|---|
| `row_id` | nonempty string | Stable unique canonical row ID. The raw exact duplicate repeats it; conflicting rows with the same ID are an error. |
| `date` | ISO `YYYY-MM-DD` | UTC source paid-click cohort date, within September 1–14. |
| `market` | enum | `UK` only; scope is paid cohorts in the fictional UK launch. |
| `channel` | enum | `google-search`, `linkedin`, `reddit-paid`. Organic Reddit activity is excluded. |
| `campaign_id` | enum after normalization | One campaign per channel; lookup below. |
| `device` | enum | `mobile` or `desktop`, device of assigned source paid click. |
| `audience_id` | enum | `skeptical` or `exploring`; mutually exclusive constructed buckets within channel, not validated customer traits. |
| `creative_id` | enum | `traceability` or `value`; message hypotheses with channel-specific format, lookup below. |
| `spend_cents` | nonnegative integer USD cents | Media cost only. No agency, creative, tooling, salaries or other acquisition costs. |
| `impressions` | nonnegative integer | Acquisition ad impressions in the bucket; not unique people or frequency. |
| `clicks` | nonnegative integer | Paid clicks assigned to this bucket; not unique visitors. |
| `landing_sessions` | nonnegative integer | Source-tagged landing sessions observed within seven days of the eligible paid click. At most one retained session per synthetic funnel journey. |
| `quiz_starts` | nonnegative integer | Distinct retained source-tagged quiz journeys that start within the same seven-day window; one retained start per journey. Completion-rate denominator. |
| `quiz_completions` | nonnegative integer | Those retained quiz journeys completed within the same seven-day window; no repeated completion counting. |
| `attributed_new_customers` | nonnegative integer | First-purchase customers completing this constructed quiz-led funnel within seven days, assigned once by the declared last-paid-click rule. Not all-site new customers. |
| `attributed_first_purchase_revenue_cents` | nonnegative integer USD cents | First-purchase revenue from those customers, after discounts, excluding tax and shipping. No refunds in this fixture; no renewal revenue. The generator assigns USD $80 (8,000 cents) per first-purchase customer for arithmetic clarity; this is a reconstruction assumption, not course pricing or LTV. |
| `is_provisional` | boolean | `false` means this row's seven-day observation window is mature at the declared snapshot. Reject unsupported string/boolean coercion. |
Per-row inequalities: `impressions >= clicks >= landing_sessions >= quiz_starts >= quiz_completions >= attributed_new_customers`. This is an explicit simplified quiz-led funnel construction, not a universal property of advertising exports. A row with zero customers must have zero first-purchase revenue; a positive customer count must have positive revenue. These constraints do not supply person-level evidence.
## Canonical lookups and raw cleanup
| Channel | Canonical campaign ID | Display name | Allowed raw campaign ID alias |
|---|---|---|---|
| `google-search` | `gl-uk-google-search` | GreenLeaf UK Search | `gl_uk_search` → `gl-uk-google-search` |
| `linkedin` | `gl-uk-linkedin` | GreenLeaf UK LinkedIn | None |
| `reddit-paid` | `gl-uk-reddit-paid` | GreenLeaf UK Reddit paid | None |
Audience `skeptical`: constructed bucket for a proof-seeking buyer comparing sourcing/dosing evidence; inspired by Oliver's unvalidated persona. Audience `exploring`: constructed bucket for a buyer exploring the category and considering value. They are neither real targeting specifications nor randomized treatment assignments; no identity, profession or health condition is inferred.
Creative `traceability`: illustrative “Know what is in every serving” message. Creative `value`: illustrative “Compare the value of a clearly dosed serving” message. For Google Search these are text-ad message hypotheses; for LinkedIn, sponsored-feed copy hypotheses; for Reddit paid, promoted-post copy hypotheses. Creative IDs are shared message categories within each channel, not identical finished ad assets or approved claims. No proof of efficacy or actual product certification is supplied.
Remove only a byte-for-byte/equal-field exact duplicate, recording its source row ID. Normalize only the declared alias and record the count. An unknown ID, unexpected market, missing field, malformed date, noninteger cents/count, negative value, conflicting duplicate or impossible funnel must fail validation rather than be silently repaired. Preserve a documented zero-denominator row to verify unavailable-ratio handling. Reconcile raw/canonical counts and all dimension totals. The fixture generation record must identify the exact duplicate, alias row and zero-denominator row; this dictionary does not pretend those rows already exist.
## Metrics and comparison rules
Use sums within the selected cohort/dimension, then divide; never average row-level rates. Retain full intermediate precision. Display USD by dividing cents by 100 only at presentation. A zero denominator yields **unavailable**, never zero/infinity; percent change against a zero prior value is unavailable.
| Metric | Definition | Interpretation |
|---|---|---|
| Quiz completion rate | `sum(quiz_completions) / sum(quiz_starts)` | Primary funnel KPI; paid-mobile Week 1/2 must be calibrated to 45%/28%. |
| Click-to-start rate | `sum(quiz_starts) / sum(clicks)` | Diagnostic; distinct from quiz completion. |
| Click-to-landing rate | `sum(landing_sessions) / sum(clicks)` | Diagnostic, not a verified bounce rate. |
| Completion-to-purchase rate | `sum(attributed_new_customers) / sum(quiz_completions)` | Constructed quiz-led funnel diagnostic. |
| CTR | `sum(clicks) / sum(impressions)` | Optimization metric; not awareness. |
| CPC (USD) | `sum(spend_cents) / sum(clicks) / 100` | Media cost per click. |
| CPM (USD) | `sum(spend_cents) / sum(impressions) * 1000 / 100` | Media cost per 1,000 impressions. |
| Media CPA (USD) | `sum(spend_cents) / sum(attributed_new_customers) / 100` | Media-only proxy; not full CAC. |
| Attributed first-order ROAS | `sum(attributed_first_purchase_revenue_cents) / sum(spend_cents)` | Revenue/spend ratio; not profit, incrementality or LTV. |
| First-purchase revenue (USD) | `sum(attributed_first_purchase_revenue_cents) / 100` | Paid-cohort attribution only; not a verdict against the $60K full-month target. |
A rate's percentage-point change is `(Week 2 rate - Week 1 rate) * 100`. Relative change is `(Week 2 rate / Week 1 rate - 1) * 100` when Week 1 is positive. The authored 45% → 28% equals −17 percentage points and approximately −37.8% relative; no other result is predetermined by this source contract.
Before diagnosing mobile aggregate movement, stratify mobile by channel × audience × creative. For stratum `s`, set fixed weight `w_s = Week 1 starts_s / Week 1 total starts`. Compare `sum(w_s * Week 1 completion_rate_s)` with `sum(w_s * Week 2 completion_rate_s)` and compare both with observed rates. Publish weights and counts. A positive-weight stratum without a defined Week 2 rate makes the full fixed-mix comparison unavailable; any explicitly restricted common-strata comparison must disclose excluded counts/weights and cannot silently stand in for the whole. A zero-weight stratum contributes zero and does not require a rate. Changing mix and within-stratum movement may coexist. Neither proves a cause; the cohorts are not randomized.
An incomplete snapshot or provisional primary row makes the two full periods unsettled/noncomparable. Do not treat unobserved outcomes as zero. A separately selected comparison with identical mature coverage must state the selected dates, exclusions and changed question.
## Unavailable evidence and source conflicts
Unavailable: full CAC, profit, retention, realized 12-month LTV, awareness, incrementality, true all-channel/month revenue, frequency, impression share, bounce rate, event/release logs, customer interviews and causal explanations. The source asset's separate 4.2% “quiz CVR” has no adequate shared denominator definition here; do not import it. No screenshot's loss estimate, segment multiple, budget instruction or organic-to-paid causal story is a derived fixture result. See the source map for course facts versus newly authored assumptions.
## Attached file: data/reconciled.csv
row_id,date,market,channel,campaign_id,device,audience_id,creative_id,spend_cents,impressions,clicks,landing_sessions,quiz_starts,quiz_completions,attributed_new_customers,attributed_first_purchase_revenue_cents,is_provisional
d01-google-search-mobile-skeptical-traceability,2026-09-01,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3100,2520,140,120,100,54,4,32000,false
d01-google-search-mobile-skeptical-value,2026-09-01,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3100,2520,140,120,100,58,4,32000,false
d01-google-search-mobile-exploring-traceability,2026-09-01,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3100,2520,140,120,100,25,2,16000,false
d01-google-search-mobile-exploring-value,2026-09-01,UK,google-search,gl-uk-google-search,mobile,exploring,value,3100,2520,140,120,100,29,2,16000,false
d01-google-search-desktop-skeptical-traceability,2026-09-01,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3100,2160,120,100,80,38,3,24000,false
d01-google-search-desktop-skeptical-value,2026-09-01,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3100,2160,120,100,80,40,3,24000,false
d01-google-search-desktop-exploring-traceability,2026-09-01,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3100,2160,120,100,80,30,2,16000,false
d01-google-search-desktop-exploring-value,2026-09-01,UK,google-search,gl-uk-google-search,desktop,exploring,value,3100,2160,120,100,80,32,2,16000,false
d01-linkedin-mobile-skeptical-traceability,2026-09-01,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4100,2660,140,120,100,58,4,32000,false
d01-linkedin-mobile-skeptical-value,2026-09-01,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4100,2660,140,120,100,62,4,32000,false
d01-linkedin-mobile-exploring-traceability,2026-09-01,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4100,2660,140,120,100,28,2,16000,false
d01-linkedin-mobile-exploring-value,2026-09-01,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4100,2660,140,120,100,32,2,16000,false
d01-linkedin-desktop-skeptical-traceability,2026-09-01,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4100,2280,120,100,80,39,3,24000,false
d01-linkedin-desktop-skeptical-value,2026-09-01,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4100,2280,120,100,80,41,3,24000,false
d01-linkedin-desktop-exploring-traceability,2026-09-01,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4100,2280,120,100,80,31,2,16000,false
d01-linkedin-desktop-exploring-value,2026-09-01,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4100,2280,120,100,80,33,2,16000,false
d01-reddit-paid-mobile-skeptical-traceability,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2600,2800,140,120,100,62,4,32000,false
d01-reddit-paid-mobile-skeptical-value,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2600,2800,140,120,100,66,5,40000,false
d01-reddit-paid-mobile-exploring-traceability,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2600,2800,140,120,100,31,2,16000,false
d01-reddit-paid-mobile-exploring-value,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2600,2800,140,120,100,35,2,16000,false
d01-reddit-paid-desktop-skeptical-traceability,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2600,2400,120,100,80,40,3,24000,false
d01-reddit-paid-desktop-skeptical-value,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2600,2400,120,100,80,42,3,24000,false
d01-reddit-paid-desktop-exploring-traceability,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2600,2400,120,100,80,32,2,16000,false
d01-reddit-paid-desktop-exploring-value,2026-09-01,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,0,0,0,0,0,0,0,0,false
d02-google-search-mobile-skeptical-traceability,2026-09-02,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3200,2520,140,120,100,54,4,32000,false
d02-google-search-mobile-skeptical-value,2026-09-02,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3200,2520,140,120,100,58,4,32000,false
d02-google-search-mobile-exploring-traceability,2026-09-02,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3200,2520,140,120,100,25,2,16000,false
d02-google-search-mobile-exploring-value,2026-09-02,UK,google-search,gl-uk-google-search,mobile,exploring,value,3200,2520,140,120,100,29,2,16000,false
d02-google-search-desktop-skeptical-traceability,2026-09-02,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3200,2160,120,100,80,38,3,24000,false
d02-google-search-desktop-skeptical-value,2026-09-02,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3200,2160,120,100,80,40,3,24000,false
d02-google-search-desktop-exploring-traceability,2026-09-02,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3200,2160,120,100,80,30,2,16000,false
d02-google-search-desktop-exploring-value,2026-09-02,UK,google-search,gl-uk-google-search,desktop,exploring,value,3200,2160,120,100,80,32,2,16000,false
d02-linkedin-mobile-skeptical-traceability,2026-09-02,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4200,2660,140,120,100,58,4,32000,false
d02-linkedin-mobile-skeptical-value,2026-09-02,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4200,2660,140,120,100,62,4,32000,false
d02-linkedin-mobile-exploring-traceability,2026-09-02,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4200,2660,140,120,100,28,2,16000,false
d02-linkedin-mobile-exploring-value,2026-09-02,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4200,2660,140,120,100,32,2,16000,false
d02-linkedin-desktop-skeptical-traceability,2026-09-02,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4200,2280,120,100,80,39,3,24000,false
d02-linkedin-desktop-skeptical-value,2026-09-02,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4200,2280,120,100,80,41,3,24000,false
d02-linkedin-desktop-exploring-traceability,2026-09-02,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4200,2280,120,100,80,31,2,16000,false
d02-linkedin-desktop-exploring-value,2026-09-02,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4200,2280,120,100,80,33,2,16000,false
d02-reddit-paid-mobile-skeptical-traceability,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2700,2800,140,120,100,62,4,32000,false
d02-reddit-paid-mobile-skeptical-value,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2700,2800,140,120,100,66,5,40000,false
d02-reddit-paid-mobile-exploring-traceability,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2700,2800,140,120,100,31,2,16000,false
d02-reddit-paid-mobile-exploring-value,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2700,2800,140,120,100,35,2,16000,false
d02-reddit-paid-desktop-skeptical-traceability,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2700,2400,120,100,80,40,3,24000,false
d02-reddit-paid-desktop-skeptical-value,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2700,2400,120,100,80,42,3,24000,false
d02-reddit-paid-desktop-exploring-traceability,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2700,2400,120,100,80,32,2,16000,false
d02-reddit-paid-desktop-exploring-value,2026-09-02,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2700,2400,120,100,80,34,2,16000,false
d03-google-search-mobile-skeptical-traceability,2026-09-03,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3000,2520,140,120,100,54,4,32000,false
d03-google-search-mobile-skeptical-value,2026-09-03,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3000,2520,140,120,100,58,4,32000,false
d03-google-search-mobile-exploring-traceability,2026-09-03,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3000,2520,140,120,100,25,2,16000,false
d03-google-search-mobile-exploring-value,2026-09-03,UK,google-search,gl-uk-google-search,mobile,exploring,value,3000,2520,140,120,100,29,2,16000,false
d03-google-search-desktop-skeptical-traceability,2026-09-03,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3000,2160,120,100,80,38,3,24000,false
d03-google-search-desktop-skeptical-value,2026-09-03,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3000,2160,120,100,80,40,3,24000,false
d03-google-search-desktop-exploring-traceability,2026-09-03,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3000,2160,120,100,80,30,2,16000,false
d03-google-search-desktop-exploring-value,2026-09-03,UK,google-search,gl-uk-google-search,desktop,exploring,value,3000,2160,120,100,80,32,2,16000,false
d03-linkedin-mobile-skeptical-traceability,2026-09-03,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4000,2660,140,120,100,58,4,32000,false
d03-linkedin-mobile-skeptical-value,2026-09-03,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4000,2660,140,120,100,62,4,32000,false
d03-linkedin-mobile-exploring-traceability,2026-09-03,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4000,2660,140,120,100,28,2,16000,false
d03-linkedin-mobile-exploring-value,2026-09-03,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4000,2660,140,120,100,32,2,16000,false
d03-linkedin-desktop-skeptical-traceability,2026-09-03,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4000,2280,120,100,80,39,3,24000,false
d03-linkedin-desktop-skeptical-value,2026-09-03,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4000,2280,120,100,80,41,3,24000,false
d03-linkedin-desktop-exploring-traceability,2026-09-03,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4000,2280,120,100,80,31,2,16000,false
d03-linkedin-desktop-exploring-value,2026-09-03,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4000,2280,120,100,80,33,2,16000,false
d03-reddit-paid-mobile-skeptical-traceability,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2500,2800,140,120,100,62,4,32000,false
d03-reddit-paid-mobile-skeptical-value,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2500,2800,140,120,100,66,5,40000,false
d03-reddit-paid-mobile-exploring-traceability,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2500,2800,140,120,100,31,2,16000,false
d03-reddit-paid-mobile-exploring-value,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2500,2800,140,120,100,35,2,16000,false
d03-reddit-paid-desktop-skeptical-traceability,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2500,2400,120,100,80,40,3,24000,false
d03-reddit-paid-desktop-skeptical-value,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2500,2400,120,100,80,42,3,24000,false
d03-reddit-paid-desktop-exploring-traceability,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2500,2400,120,100,80,32,2,16000,false
d03-reddit-paid-desktop-exploring-value,2026-09-03,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2500,2400,120,100,80,34,2,16000,false
d04-google-search-mobile-skeptical-traceability,2026-09-04,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3100,2520,140,120,100,54,4,32000,false
d04-google-search-mobile-skeptical-value,2026-09-04,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3100,2520,140,120,100,58,4,32000,false
d04-google-search-mobile-exploring-traceability,2026-09-04,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3100,2520,140,120,100,25,2,16000,false
d04-google-search-mobile-exploring-value,2026-09-04,UK,google-search,gl-uk-google-search,mobile,exploring,value,3100,2520,140,120,100,29,2,16000,false
d04-google-search-desktop-skeptical-traceability,2026-09-04,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3100,2160,120,100,80,38,3,24000,false
d04-google-search-desktop-skeptical-value,2026-09-04,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3100,2160,120,100,80,40,3,24000,false
d04-google-search-desktop-exploring-traceability,2026-09-04,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3100,2160,120,100,80,30,2,16000,false
d04-google-search-desktop-exploring-value,2026-09-04,UK,google-search,gl-uk-google-search,desktop,exploring,value,3100,2160,120,100,80,32,2,16000,false
d04-linkedin-mobile-skeptical-traceability,2026-09-04,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4100,2660,140,120,100,58,4,32000,false
d04-linkedin-mobile-skeptical-value,2026-09-04,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4100,2660,140,120,100,62,4,32000,false
d04-linkedin-mobile-exploring-traceability,2026-09-04,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4100,2660,140,120,100,28,2,16000,false
d04-linkedin-mobile-exploring-value,2026-09-04,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4100,2660,140,120,100,32,2,16000,false
d04-linkedin-desktop-skeptical-traceability,2026-09-04,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4100,2280,120,100,80,39,3,24000,false
d04-linkedin-desktop-skeptical-value,2026-09-04,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4100,2280,120,100,80,41,3,24000,false
d04-linkedin-desktop-exploring-traceability,2026-09-04,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4100,2280,120,100,80,31,2,16000,false
d04-linkedin-desktop-exploring-value,2026-09-04,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4100,2280,120,100,80,33,2,16000,false
d04-reddit-paid-mobile-skeptical-traceability,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2600,2800,140,120,100,62,4,32000,false
d04-reddit-paid-mobile-skeptical-value,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2600,2800,140,120,100,66,5,40000,false
d04-reddit-paid-mobile-exploring-traceability,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2600,2800,140,120,100,31,2,16000,false
d04-reddit-paid-mobile-exploring-value,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2600,2800,140,120,100,35,2,16000,false
d04-reddit-paid-desktop-skeptical-traceability,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2600,2400,120,100,80,40,3,24000,false
d04-reddit-paid-desktop-skeptical-value,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2600,2400,120,100,80,42,3,24000,false
d04-reddit-paid-desktop-exploring-traceability,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2600,2400,120,100,80,32,2,16000,false
d04-reddit-paid-desktop-exploring-value,2026-09-04,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2600,2400,120,100,80,34,2,16000,false
d05-google-search-mobile-skeptical-traceability,2026-09-05,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3200,2520,140,120,100,54,4,32000,false
d05-google-search-mobile-skeptical-value,2026-09-05,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3200,2520,140,120,100,58,4,32000,false
d05-google-search-mobile-exploring-traceability,2026-09-05,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3200,2520,140,120,100,25,2,16000,false
d05-google-search-mobile-exploring-value,2026-09-05,UK,google-search,gl-uk-google-search,mobile,exploring,value,3200,2520,140,120,100,29,2,16000,false
d05-google-search-desktop-skeptical-traceability,2026-09-05,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3200,2160,120,100,80,38,3,24000,false
d05-google-search-desktop-skeptical-value,2026-09-05,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3200,2160,120,100,80,40,3,24000,false
d05-google-search-desktop-exploring-traceability,2026-09-05,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3200,2160,120,100,80,30,2,16000,false
d05-google-search-desktop-exploring-value,2026-09-05,UK,google-search,gl-uk-google-search,desktop,exploring,value,3200,2160,120,100,80,32,2,16000,false
d05-linkedin-mobile-skeptical-traceability,2026-09-05,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4200,2660,140,120,100,58,4,32000,false
d05-linkedin-mobile-skeptical-value,2026-09-05,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4200,2660,140,120,100,62,4,32000,false
d05-linkedin-mobile-exploring-traceability,2026-09-05,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4200,2660,140,120,100,28,2,16000,false
d05-linkedin-mobile-exploring-value,2026-09-05,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4200,2660,140,120,100,32,2,16000,false
d05-linkedin-desktop-skeptical-traceability,2026-09-05,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4200,2280,120,100,80,39,3,24000,false
d05-linkedin-desktop-skeptical-value,2026-09-05,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4200,2280,120,100,80,41,3,24000,false
d05-linkedin-desktop-exploring-traceability,2026-09-05,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4200,2280,120,100,80,31,2,16000,false
d05-linkedin-desktop-exploring-value,2026-09-05,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4200,2280,120,100,80,33,2,16000,false
d05-reddit-paid-mobile-skeptical-traceability,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2700,2800,140,120,100,62,4,32000,false
d05-reddit-paid-mobile-skeptical-value,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2700,2800,140,120,100,66,5,40000,false
d05-reddit-paid-mobile-exploring-traceability,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2700,2800,140,120,100,31,2,16000,false
d05-reddit-paid-mobile-exploring-value,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2700,2800,140,120,100,35,2,16000,false
d05-reddit-paid-desktop-skeptical-traceability,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2700,2400,120,100,80,40,3,24000,false
d05-reddit-paid-desktop-skeptical-value,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2700,2400,120,100,80,42,3,24000,false
d05-reddit-paid-desktop-exploring-traceability,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2700,2400,120,100,80,32,2,16000,false
d05-reddit-paid-desktop-exploring-value,2026-09-05,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2700,2400,120,100,80,34,2,16000,false
d06-google-search-mobile-skeptical-traceability,2026-09-06,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3000,2520,140,120,100,54,4,32000,false
d06-google-search-mobile-skeptical-value,2026-09-06,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3000,2520,140,120,100,58,4,32000,false
d06-google-search-mobile-exploring-traceability,2026-09-06,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3000,2520,140,120,100,25,2,16000,false
d06-google-search-mobile-exploring-value,2026-09-06,UK,google-search,gl-uk-google-search,mobile,exploring,value,3000,2520,140,120,100,29,2,16000,false
d06-google-search-desktop-skeptical-traceability,2026-09-06,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3000,2160,120,100,80,38,3,24000,false
d06-google-search-desktop-skeptical-value,2026-09-06,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3000,2160,120,100,80,40,3,24000,false
d06-google-search-desktop-exploring-traceability,2026-09-06,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3000,2160,120,100,80,30,2,16000,false
d06-google-search-desktop-exploring-value,2026-09-06,UK,google-search,gl-uk-google-search,desktop,exploring,value,3000,2160,120,100,80,32,2,16000,false
d06-linkedin-mobile-skeptical-traceability,2026-09-06,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4000,2660,140,120,100,58,4,32000,false
d06-linkedin-mobile-skeptical-value,2026-09-06,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4000,2660,140,120,100,62,4,32000,false
d06-linkedin-mobile-exploring-traceability,2026-09-06,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4000,2660,140,120,100,28,2,16000,false
d06-linkedin-mobile-exploring-value,2026-09-06,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4000,2660,140,120,100,32,2,16000,false
d06-linkedin-desktop-skeptical-traceability,2026-09-06,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4000,2280,120,100,80,39,3,24000,false
d06-linkedin-desktop-skeptical-value,2026-09-06,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4000,2280,120,100,80,41,3,24000,false
d06-linkedin-desktop-exploring-traceability,2026-09-06,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4000,2280,120,100,80,31,2,16000,false
d06-linkedin-desktop-exploring-value,2026-09-06,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4000,2280,120,100,80,33,2,16000,false
d06-reddit-paid-mobile-skeptical-traceability,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2500,2800,140,120,100,62,4,32000,false
d06-reddit-paid-mobile-skeptical-value,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2500,2800,140,120,100,66,5,40000,false
d06-reddit-paid-mobile-exploring-traceability,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2500,2800,140,120,100,31,2,16000,false
d06-reddit-paid-mobile-exploring-value,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2500,2800,140,120,100,35,2,16000,false
d06-reddit-paid-desktop-skeptical-traceability,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2500,2400,120,100,80,40,3,24000,false
d06-reddit-paid-desktop-skeptical-value,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2500,2400,120,100,80,42,3,24000,false
d06-reddit-paid-desktop-exploring-traceability,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2500,2400,120,100,80,32,2,16000,false
d06-reddit-paid-desktop-exploring-value,2026-09-06,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2500,2400,120,100,80,34,2,16000,false
d07-google-search-mobile-skeptical-traceability,2026-09-07,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3100,2520,140,120,100,54,4,32000,false
d07-google-search-mobile-skeptical-value,2026-09-07,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3100,2520,140,120,100,58,4,32000,false
d07-google-search-mobile-exploring-traceability,2026-09-07,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3100,2520,140,120,100,25,2,16000,false
d07-google-search-mobile-exploring-value,2026-09-07,UK,google-search,gl-uk-google-search,mobile,exploring,value,3100,2520,140,120,100,29,2,16000,false
d07-google-search-desktop-skeptical-traceability,2026-09-07,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3100,2160,120,100,80,38,3,24000,false
d07-google-search-desktop-skeptical-value,2026-09-07,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3100,2160,120,100,80,40,3,24000,false
d07-google-search-desktop-exploring-traceability,2026-09-07,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3100,2160,120,100,80,30,2,16000,false
d07-google-search-desktop-exploring-value,2026-09-07,UK,google-search,gl-uk-google-search,desktop,exploring,value,3100,2160,120,100,80,32,2,16000,false
d07-linkedin-mobile-skeptical-traceability,2026-09-07,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4100,2660,140,120,100,58,4,32000,false
d07-linkedin-mobile-skeptical-value,2026-09-07,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4100,2660,140,120,100,62,4,32000,false
d07-linkedin-mobile-exploring-traceability,2026-09-07,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4100,2660,140,120,100,28,2,16000,false
d07-linkedin-mobile-exploring-value,2026-09-07,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4100,2660,140,120,100,32,2,16000,false
d07-linkedin-desktop-skeptical-traceability,2026-09-07,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4100,2280,120,100,80,39,3,24000,false
d07-linkedin-desktop-skeptical-value,2026-09-07,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4100,2280,120,100,80,41,3,24000,false
d07-linkedin-desktop-exploring-traceability,2026-09-07,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4100,2280,120,100,80,31,2,16000,false
d07-linkedin-desktop-exploring-value,2026-09-07,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4100,2280,120,100,80,33,2,16000,false
d07-reddit-paid-mobile-skeptical-traceability,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2600,2800,140,120,100,62,4,32000,false
d07-reddit-paid-mobile-skeptical-value,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2600,2800,140,120,100,66,5,40000,false
d07-reddit-paid-mobile-exploring-traceability,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2600,2800,140,120,100,31,2,16000,false
d07-reddit-paid-mobile-exploring-value,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2600,2800,140,120,100,35,2,16000,false
d07-reddit-paid-desktop-skeptical-traceability,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2600,2400,120,100,80,40,3,24000,false
d07-reddit-paid-desktop-skeptical-value,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2600,2400,120,100,80,42,3,24000,false
d07-reddit-paid-desktop-exploring-traceability,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2600,2400,120,100,80,32,2,16000,false
d07-reddit-paid-desktop-exploring-value,2026-09-07,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2600,2400,120,100,80,34,2,16000,false
d08-google-search-mobile-skeptical-traceability,2026-09-08,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3200,1620,90,70,50,22,1,8000,false
d08-google-search-mobile-skeptical-value,2026-09-08,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3200,1620,90,70,50,24,1,8000,false
d08-google-search-mobile-exploring-traceability,2026-09-08,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3200,3420,190,170,150,26,2,16000,false
d08-google-search-mobile-exploring-value,2026-09-08,UK,google-search,gl-uk-google-search,mobile,exploring,value,3200,3420,190,170,150,30,2,16000,false
d08-google-search-desktop-skeptical-traceability,2026-09-08,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3200,2520,140,120,100,47,3,24000,false
d08-google-search-desktop-skeptical-value,2026-09-08,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3200,2520,140,120,100,49,3,24000,false
d08-google-search-desktop-exploring-traceability,2026-09-08,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3200,1800,100,80,60,23,1,8000,false
d08-google-search-desktop-exploring-value,2026-09-08,UK,google-search,gl-uk-google-search,desktop,exploring,value,3200,1800,100,80,60,25,2,16000,false
d08-linkedin-mobile-skeptical-traceability,2026-09-08,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4200,1710,90,70,50,24,1,8000,false
d08-linkedin-mobile-skeptical-value,2026-09-08,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4200,1710,90,70,50,26,2,16000,false
d08-linkedin-mobile-exploring-traceability,2026-09-08,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4200,3610,190,170,150,29,2,16000,false
d08-linkedin-mobile-exploring-value,2026-09-08,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4200,3610,190,170,150,33,2,16000,false
d08-linkedin-desktop-skeptical-traceability,2026-09-08,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4200,2660,140,120,100,48,3,24000,false
d08-linkedin-desktop-skeptical-value,2026-09-08,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4200,2660,140,120,100,50,4,32000,false
d08-linkedin-desktop-exploring-traceability,2026-09-08,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4200,1900,100,80,60,24,1,8000,false
d08-linkedin-desktop-exploring-value,2026-09-08,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4200,1900,100,80,60,26,2,16000,false
d08-reddit-paid-mobile-skeptical-traceability,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2700,1800,90,70,50,26,2,16000,false
d08-reddit-paid-mobile-skeptical-value,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2700,1800,90,70,50,28,2,16000,false
d08-reddit-paid-mobile-exploring-traceability,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2700,3800,190,170,150,32,2,16000,false
d08-reddit-paid-mobile-exploring-value,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2700,3800,190,170,150,36,2,16000,false
d08-reddit-paid-desktop-skeptical-traceability,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2700,2800,140,120,100,49,3,24000,false
d08-reddit-paid-desktop-skeptical-value,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2700,2800,140,120,100,51,4,32000,false
d08-reddit-paid-desktop-exploring-traceability,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2700,2000,100,80,60,25,2,16000,false
d08-reddit-paid-desktop-exploring-value,2026-09-08,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2700,2000,100,80,60,27,2,16000,false
d09-google-search-mobile-skeptical-traceability,2026-09-09,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3000,1620,90,70,50,22,1,8000,false
d09-google-search-mobile-skeptical-value,2026-09-09,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3000,1620,90,70,50,24,1,8000,false
d09-google-search-mobile-exploring-traceability,2026-09-09,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3000,3420,190,170,150,26,2,16000,false
d09-google-search-mobile-exploring-value,2026-09-09,UK,google-search,gl-uk-google-search,mobile,exploring,value,3000,3420,190,170,150,30,2,16000,false
d09-google-search-desktop-skeptical-traceability,2026-09-09,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3000,2520,140,120,100,47,3,24000,false
d09-google-search-desktop-skeptical-value,2026-09-09,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3000,2520,140,120,100,49,3,24000,false
d09-google-search-desktop-exploring-traceability,2026-09-09,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3000,1800,100,80,60,23,1,8000,false
d09-google-search-desktop-exploring-value,2026-09-09,UK,google-search,gl-uk-google-search,desktop,exploring,value,3000,1800,100,80,60,25,2,16000,false
d09-linkedin-mobile-skeptical-traceability,2026-09-09,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4000,1710,90,70,50,24,1,8000,false
d09-linkedin-mobile-skeptical-value,2026-09-09,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4000,1710,90,70,50,26,2,16000,false
d09-linkedin-mobile-exploring-traceability,2026-09-09,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4000,3610,190,170,150,29,2,16000,false
d09-linkedin-mobile-exploring-value,2026-09-09,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4000,3610,190,170,150,33,2,16000,false
d09-linkedin-desktop-skeptical-traceability,2026-09-09,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4000,2660,140,120,100,48,3,24000,false
d09-linkedin-desktop-skeptical-value,2026-09-09,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4000,2660,140,120,100,50,4,32000,false
d09-linkedin-desktop-exploring-traceability,2026-09-09,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4000,1900,100,80,60,24,1,8000,false
d09-linkedin-desktop-exploring-value,2026-09-09,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4000,1900,100,80,60,26,2,16000,false
d09-reddit-paid-mobile-skeptical-traceability,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2500,1800,90,70,50,26,2,16000,false
d09-reddit-paid-mobile-skeptical-value,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2500,1800,90,70,50,28,2,16000,false
d09-reddit-paid-mobile-exploring-traceability,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2500,3800,190,170,150,32,2,16000,false
d09-reddit-paid-mobile-exploring-value,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2500,3800,190,170,150,36,2,16000,false
d09-reddit-paid-desktop-skeptical-traceability,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2500,2800,140,120,100,49,3,24000,false
d09-reddit-paid-desktop-skeptical-value,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2500,2800,140,120,100,51,4,32000,false
d09-reddit-paid-desktop-exploring-traceability,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2500,2000,100,80,60,25,2,16000,false
d09-reddit-paid-desktop-exploring-value,2026-09-09,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2500,2000,100,80,60,27,2,16000,false
d10-google-search-mobile-skeptical-traceability,2026-09-10,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3100,1620,90,70,50,22,1,8000,false
d10-google-search-mobile-skeptical-value,2026-09-10,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3100,1620,90,70,50,24,1,8000,false
d10-google-search-mobile-exploring-traceability,2026-09-10,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3100,3420,190,170,150,26,2,16000,false
d10-google-search-mobile-exploring-value,2026-09-10,UK,google-search,gl-uk-google-search,mobile,exploring,value,3100,3420,190,170,150,30,2,16000,false
d10-google-search-desktop-skeptical-traceability,2026-09-10,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3100,2520,140,120,100,47,3,24000,false
d10-google-search-desktop-skeptical-value,2026-09-10,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3100,2520,140,120,100,49,3,24000,false
d10-google-search-desktop-exploring-traceability,2026-09-10,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3100,1800,100,80,60,23,1,8000,false
d10-google-search-desktop-exploring-value,2026-09-10,UK,google-search,gl-uk-google-search,desktop,exploring,value,3100,1800,100,80,60,25,2,16000,false
d10-linkedin-mobile-skeptical-traceability,2026-09-10,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4100,1710,90,70,50,24,1,8000,false
d10-linkedin-mobile-skeptical-value,2026-09-10,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4100,1710,90,70,50,26,2,16000,false
d10-linkedin-mobile-exploring-traceability,2026-09-10,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4100,3610,190,170,150,29,2,16000,false
d10-linkedin-mobile-exploring-value,2026-09-10,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4100,3610,190,170,150,33,2,16000,false
d10-linkedin-desktop-skeptical-traceability,2026-09-10,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4100,2660,140,120,100,48,3,24000,false
d10-linkedin-desktop-skeptical-value,2026-09-10,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4100,2660,140,120,100,50,4,32000,false
d10-linkedin-desktop-exploring-traceability,2026-09-10,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4100,1900,100,80,60,24,1,8000,false
d10-linkedin-desktop-exploring-value,2026-09-10,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4100,1900,100,80,60,26,2,16000,false
d10-reddit-paid-mobile-skeptical-traceability,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2600,1800,90,70,50,26,2,16000,false
d10-reddit-paid-mobile-skeptical-value,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2600,1800,90,70,50,28,2,16000,false
d10-reddit-paid-mobile-exploring-traceability,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2600,3800,190,170,150,32,2,16000,false
d10-reddit-paid-mobile-exploring-value,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2600,3800,190,170,150,36,2,16000,false
d10-reddit-paid-desktop-skeptical-traceability,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2600,2800,140,120,100,49,3,24000,false
d10-reddit-paid-desktop-skeptical-value,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2600,2800,140,120,100,51,4,32000,false
d10-reddit-paid-desktop-exploring-traceability,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2600,2000,100,80,60,25,2,16000,false
d10-reddit-paid-desktop-exploring-value,2026-09-10,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2600,2000,100,80,60,27,2,16000,false
d11-google-search-mobile-skeptical-traceability,2026-09-11,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3200,1620,90,70,50,22,1,8000,false
d11-google-search-mobile-skeptical-value,2026-09-11,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3200,1620,90,70,50,24,1,8000,false
d11-google-search-mobile-exploring-traceability,2026-09-11,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3200,3420,190,170,150,26,2,16000,false
d11-google-search-mobile-exploring-value,2026-09-11,UK,google-search,gl-uk-google-search,mobile,exploring,value,3200,3420,190,170,150,30,2,16000,false
d11-google-search-desktop-skeptical-traceability,2026-09-11,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3200,2520,140,120,100,47,3,24000,false
d11-google-search-desktop-skeptical-value,2026-09-11,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3200,2520,140,120,100,49,3,24000,false
d11-google-search-desktop-exploring-traceability,2026-09-11,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3200,1800,100,80,60,23,1,8000,false
d11-google-search-desktop-exploring-value,2026-09-11,UK,google-search,gl-uk-google-search,desktop,exploring,value,3200,1800,100,80,60,25,2,16000,false
d11-linkedin-mobile-skeptical-traceability,2026-09-11,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4200,1710,90,70,50,24,1,8000,false
d11-linkedin-mobile-skeptical-value,2026-09-11,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4200,1710,90,70,50,26,2,16000,false
d11-linkedin-mobile-exploring-traceability,2026-09-11,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4200,3610,190,170,150,29,2,16000,false
d11-linkedin-mobile-exploring-value,2026-09-11,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4200,3610,190,170,150,33,2,16000,false
d11-linkedin-desktop-skeptical-traceability,2026-09-11,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4200,2660,140,120,100,48,3,24000,false
d11-linkedin-desktop-skeptical-value,2026-09-11,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4200,2660,140,120,100,50,4,32000,false
d11-linkedin-desktop-exploring-traceability,2026-09-11,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4200,1900,100,80,60,24,1,8000,false
d11-linkedin-desktop-exploring-value,2026-09-11,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4200,1900,100,80,60,26,2,16000,false
d11-reddit-paid-mobile-skeptical-traceability,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2700,1800,90,70,50,26,2,16000,false
d11-reddit-paid-mobile-skeptical-value,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2700,1800,90,70,50,28,2,16000,false
d11-reddit-paid-mobile-exploring-traceability,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2700,3800,190,170,150,32,2,16000,false
d11-reddit-paid-mobile-exploring-value,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2700,3800,190,170,150,36,2,16000,false
d11-reddit-paid-desktop-skeptical-traceability,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2700,2800,140,120,100,49,3,24000,false
d11-reddit-paid-desktop-skeptical-value,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2700,2800,140,120,100,51,4,32000,false
d11-reddit-paid-desktop-exploring-traceability,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2700,2000,100,80,60,25,2,16000,false
d11-reddit-paid-desktop-exploring-value,2026-09-11,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2700,2000,100,80,60,27,2,16000,false
d12-google-search-mobile-skeptical-traceability,2026-09-12,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3000,1620,90,70,50,22,1,8000,false
d12-google-search-mobile-skeptical-value,2026-09-12,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3000,1620,90,70,50,24,1,8000,false
d12-google-search-mobile-exploring-traceability,2026-09-12,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3000,3420,190,170,150,26,2,16000,false
d12-google-search-mobile-exploring-value,2026-09-12,UK,google-search,gl-uk-google-search,mobile,exploring,value,3000,3420,190,170,150,30,2,16000,false
d12-google-search-desktop-skeptical-traceability,2026-09-12,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3000,2520,140,120,100,47,3,24000,false
d12-google-search-desktop-skeptical-value,2026-09-12,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3000,2520,140,120,100,49,3,24000,false
d12-google-search-desktop-exploring-traceability,2026-09-12,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3000,1800,100,80,60,23,1,8000,false
d12-google-search-desktop-exploring-value,2026-09-12,UK,google-search,gl-uk-google-search,desktop,exploring,value,3000,1800,100,80,60,25,2,16000,false
d12-linkedin-mobile-skeptical-traceability,2026-09-12,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4000,1710,90,70,50,24,1,8000,false
d12-linkedin-mobile-skeptical-value,2026-09-12,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4000,1710,90,70,50,26,2,16000,false
d12-linkedin-mobile-exploring-traceability,2026-09-12,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4000,3610,190,170,150,29,2,16000,false
d12-linkedin-mobile-exploring-value,2026-09-12,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4000,3610,190,170,150,33,2,16000,false
d12-linkedin-desktop-skeptical-traceability,2026-09-12,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4000,2660,140,120,100,48,3,24000,false
d12-linkedin-desktop-skeptical-value,2026-09-12,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4000,2660,140,120,100,50,4,32000,false
d12-linkedin-desktop-exploring-traceability,2026-09-12,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4000,1900,100,80,60,24,1,8000,false
d12-linkedin-desktop-exploring-value,2026-09-12,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4000,1900,100,80,60,26,2,16000,false
d12-reddit-paid-mobile-skeptical-traceability,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2500,1800,90,70,50,26,2,16000,false
d12-reddit-paid-mobile-skeptical-value,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2500,1800,90,70,50,28,2,16000,false
d12-reddit-paid-mobile-exploring-traceability,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2500,3800,190,170,150,32,2,16000,false
d12-reddit-paid-mobile-exploring-value,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2500,3800,190,170,150,36,2,16000,false
d12-reddit-paid-desktop-skeptical-traceability,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2500,2800,140,120,100,49,3,24000,false
d12-reddit-paid-desktop-skeptical-value,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2500,2800,140,120,100,51,4,32000,false
d12-reddit-paid-desktop-exploring-traceability,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2500,2000,100,80,60,25,2,16000,false
d12-reddit-paid-desktop-exploring-value,2026-09-12,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2500,2000,100,80,60,27,2,16000,false
d13-google-search-mobile-skeptical-traceability,2026-09-13,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3100,1620,90,70,50,22,1,8000,false
d13-google-search-mobile-skeptical-value,2026-09-13,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3100,1620,90,70,50,24,1,8000,false
d13-google-search-mobile-exploring-traceability,2026-09-13,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3100,3420,190,170,150,26,2,16000,false
d13-google-search-mobile-exploring-value,2026-09-13,UK,google-search,gl-uk-google-search,mobile,exploring,value,3100,3420,190,170,150,30,2,16000,false
d13-google-search-desktop-skeptical-traceability,2026-09-13,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3100,2520,140,120,100,47,3,24000,false
d13-google-search-desktop-skeptical-value,2026-09-13,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3100,2520,140,120,100,49,3,24000,false
d13-google-search-desktop-exploring-traceability,2026-09-13,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3100,1800,100,80,60,23,1,8000,false
d13-google-search-desktop-exploring-value,2026-09-13,UK,google-search,gl-uk-google-search,desktop,exploring,value,3100,1800,100,80,60,25,2,16000,false
d13-linkedin-mobile-skeptical-traceability,2026-09-13,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4100,1710,90,70,50,24,1,8000,false
d13-linkedin-mobile-skeptical-value,2026-09-13,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4100,1710,90,70,50,26,2,16000,false
d13-linkedin-mobile-exploring-traceability,2026-09-13,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4100,3610,190,170,150,29,2,16000,false
d13-linkedin-mobile-exploring-value,2026-09-13,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4100,3610,190,170,150,33,2,16000,false
d13-linkedin-desktop-skeptical-traceability,2026-09-13,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4100,2660,140,120,100,48,3,24000,false
d13-linkedin-desktop-skeptical-value,2026-09-13,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4100,2660,140,120,100,50,4,32000,false
d13-linkedin-desktop-exploring-traceability,2026-09-13,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4100,1900,100,80,60,24,1,8000,false
d13-linkedin-desktop-exploring-value,2026-09-13,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4100,1900,100,80,60,26,2,16000,false
d13-reddit-paid-mobile-skeptical-traceability,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2600,1800,90,70,50,26,2,16000,false
d13-reddit-paid-mobile-skeptical-value,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2600,1800,90,70,50,28,2,16000,false
d13-reddit-paid-mobile-exploring-traceability,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2600,3800,190,170,150,32,2,16000,false
d13-reddit-paid-mobile-exploring-value,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2600,3800,190,170,150,36,2,16000,false
d13-reddit-paid-desktop-skeptical-traceability,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2600,2800,140,120,100,49,3,24000,false
d13-reddit-paid-desktop-skeptical-value,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2600,2800,140,120,100,51,4,32000,false
d13-reddit-paid-desktop-exploring-traceability,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2600,2000,100,80,60,25,2,16000,false
d13-reddit-paid-desktop-exploring-value,2026-09-13,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2600,2000,100,80,60,27,2,16000,false
d14-google-search-mobile-skeptical-traceability,2026-09-14,UK,google-search,gl-uk-google-search,mobile,skeptical,traceability,3200,1620,90,70,50,22,1,8000,false
d14-google-search-mobile-skeptical-value,2026-09-14,UK,google-search,gl-uk-google-search,mobile,skeptical,value,3200,1620,90,70,50,24,1,8000,false
d14-google-search-mobile-exploring-traceability,2026-09-14,UK,google-search,gl-uk-google-search,mobile,exploring,traceability,3200,3420,190,170,150,26,2,16000,false
d14-google-search-mobile-exploring-value,2026-09-14,UK,google-search,gl-uk-google-search,mobile,exploring,value,3200,3420,190,170,150,30,2,16000,false
d14-google-search-desktop-skeptical-traceability,2026-09-14,UK,google-search,gl-uk-google-search,desktop,skeptical,traceability,3200,2520,140,120,100,47,3,24000,false
d14-google-search-desktop-skeptical-value,2026-09-14,UK,google-search,gl-uk-google-search,desktop,skeptical,value,3200,2520,140,120,100,49,3,24000,false
d14-google-search-desktop-exploring-traceability,2026-09-14,UK,google-search,gl-uk-google-search,desktop,exploring,traceability,3200,1800,100,80,60,23,1,8000,false
d14-google-search-desktop-exploring-value,2026-09-14,UK,google-search,gl-uk-google-search,desktop,exploring,value,3200,1800,100,80,60,25,2,16000,false
d14-linkedin-mobile-skeptical-traceability,2026-09-14,UK,linkedin,gl-uk-linkedin,mobile,skeptical,traceability,4200,1710,90,70,50,24,1,8000,false
d14-linkedin-mobile-skeptical-value,2026-09-14,UK,linkedin,gl-uk-linkedin,mobile,skeptical,value,4200,1710,90,70,50,26,2,16000,false
d14-linkedin-mobile-exploring-traceability,2026-09-14,UK,linkedin,gl-uk-linkedin,mobile,exploring,traceability,4200,3610,190,170,150,29,2,16000,false
d14-linkedin-mobile-exploring-value,2026-09-14,UK,linkedin,gl-uk-linkedin,mobile,exploring,value,4200,3610,190,170,150,33,2,16000,false
d14-linkedin-desktop-skeptical-traceability,2026-09-14,UK,linkedin,gl-uk-linkedin,desktop,skeptical,traceability,4200,2660,140,120,100,48,3,24000,false
d14-linkedin-desktop-skeptical-value,2026-09-14,UK,linkedin,gl-uk-linkedin,desktop,skeptical,value,4200,2660,140,120,100,50,4,32000,false
d14-linkedin-desktop-exploring-traceability,2026-09-14,UK,linkedin,gl-uk-linkedin,desktop,exploring,traceability,4200,1900,100,80,60,24,1,8000,false
d14-linkedin-desktop-exploring-value,2026-09-14,UK,linkedin,gl-uk-linkedin,desktop,exploring,value,4200,1900,100,80,60,26,2,16000,false
d14-reddit-paid-mobile-skeptical-traceability,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,traceability,2700,1800,90,70,50,26,2,16000,false
d14-reddit-paid-mobile-skeptical-value,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,mobile,skeptical,value,2700,1800,90,70,50,28,2,16000,false
d14-reddit-paid-mobile-exploring-traceability,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,traceability,2700,3800,190,170,150,32,2,16000,false
d14-reddit-paid-mobile-exploring-value,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,mobile,exploring,value,2700,3800,190,170,150,36,2,16000,false
d14-reddit-paid-desktop-skeptical-traceability,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,traceability,2700,2800,140,120,100,49,3,24000,false
d14-reddit-paid-desktop-skeptical-value,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,desktop,skeptical,value,2700,2800,140,120,100,51,4,32000,false
d14-reddit-paid-desktop-exploring-traceability,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,traceability,2700,2000,100,80,60,25,2,16000,false
d14-reddit-paid-desktop-exploring-value,2026-09-14,UK,reddit-paid,gl-uk-reddit-paid,desktop,exploring,value,2700,2000,100,80,60,27,2,16000,false
## Attached file: data/campaigns.csv
campaign_id,campaign_name,display_name,channel
gl-uk-google-search,GreenLeaf_GoogleSearch_UK_Quiz_202609,GreenLeaf UK Search,google-search
gl-uk-linkedin,GreenLeaf_LinkedIn_UK_Quiz_202609,GreenLeaf UK LinkedIn,linkedin
gl-uk-reddit-paid,GreenLeaf_RedditPaid_UK_Quiz_202609,GreenLeaf UK Reddit paid,reddit-paid
## Attached file: data/creatives.csv
channel,creative_id,format,message,status
google-search,traceability,text-ad,Know what is in every serving,illustrative-unvalidated-hypothesis
google-search,value,text-ad,Compare the value of a clearly dosed serving,illustrative-unvalidated-hypothesis
linkedin,traceability,sponsored-feed,Know what is in every serving,illustrative-unvalidated-hypothesis
linkedin,value,sponsored-feed,Compare the value of a clearly dosed serving,illustrative-unvalidated-hypothesis
reddit-paid,traceability,promoted-post,Know what is in every serving,illustrative-unvalidated-hypothesis
reddit-paid,value,promoted-post,Compare the value of a clearly dosed serving,illustrative-unvalidated-hypothesis
Follow-up response
# Mobile Composition Follow-Up
## 1. Stratum-by-stratum table
All 12 mobile strata use the channel × audience × creative dimensions from the dictionary. Week 1 and Week 2 counts are summed across the seven days in each period.
| # | Channel | Audience | Creative | W1 Starts | W1 Comps | W1 Rate | W2 Starts | W2 Comps | W2 Rate | W1 Weight |
|---:|---|---|---|---:|---:|---:|---:|---:|---:|---:|
| 1 | Google Search | Skeptical | Traceability | 700 | 378 | 54.0% | 350 | 154 | 44.0% | 0.0833 |
| 2 | Google Search | Skeptical | Value | 700 | 406 | 58.0% | 350 | 168 | 48.0% | 0.0833 |
| 3 | Google Search | Exploring | Traceability | 700 | 175 | 25.0% | 1,050 | 182 | 17.3% | 0.0833 |
| 4 | Google Search | Exploring | Value | 700 | 203 | 29.0% | 1,050 | 210 | 20.0% | 0.0833 |
| 5 | LinkedIn | Skeptical | Traceability | 700 | 406 | 58.0% | 350 | 168 | 48.0% | 0.0833 |
| 6 | LinkedIn | Skeptical | Value | 700 | 434 | 62.0% | 350 | 182 | 52.0% | 0.0833 |
| 7 | LinkedIn | Exploring | Traceability | 700 | 196 | 28.0% | 1,050 | 203 | 19.3% | 0.0833 |
| 8 | LinkedIn | Exploring | Value | 700 | 224 | 32.0% | 1,050 | 231 | 22.0% | 0.0833 |
| 9 | Reddit Paid | Skeptical | Traceability | 700 | 434 | 62.0% | 350 | 182 | 52.0% | 0.0833 |
| 10 | Reddit Paid | Skeptical | Value | 700 | 462 | 66.0% | 350 | 196 | 56.0% | 0.0833 |
| 11 | Reddit Paid | Exploring | Traceability | 700 | 217 | 31.0% | 1,050 | 224 | 21.3% | 0.0833 |
| 12 | Reddit Paid | Exploring | Value | 700 | 245 | 35.0% | 1,050 | 245 | 23.3% | 0.0833 |
| **Total** | | | | **8,400** | **3,780** | **45.0%** | **8,400** | **2,360** | **28.1%** | **1.0000** |
### Arithmetic checks
Week 1 totals:
```text
W1 starts = 12 × 700 = 8,400
W1 comps = 378 + 406 + 175 + 203 + 406 + 434 + 196 + 224 + 434 + 462 + 217 + 245
= 3,780
W1 rate = 3,780 / 8,400 = 0.45 = 45.0%
```
Week 2 starts are not uniform at 700 per stratum. The CSV pattern gives:
- Skeptical mobile strata: 50 starts per day × 7 days = **350**
- Exploring mobile traceability and value strata: 150 starts per day × 7 days = **1,050**
```text
W2 skeptical starts = 6 strata × 350 = 2,100
W2 exploring starts = 6 strata × 1,050 = 6,300
W2 starts = 2,100 + 6,300 = 8,400
```
This matches the dictionary’s aggregate mobile start total for Week 2. The rates in the table are calculated from these counts, e.g.:
```text
Google / skeptical / traceability W2:
154 / 350 = 0.4400 = 44.0%
Google / exploring / traceability W2:
182 / 1,050 = 0.17333… = 17.3%
```
## 2. Week 1 weights
Week 1 weights are each stratum’s Week 1 starts divided by total Week 1 mobile starts.
Because every Week 1 stratum has 700 starts and total Week 1 starts are 8,400:
```text
weight for each stratum = 700 / 8,400 = 1 / 12 = 0.083333…
```
Sum of weights:
```text
12 × (1 / 12) = 1.0000
```
No zero-weight or missing-weight stratum is present.
## 3. Weighted Week 2 contribution
The fixed-mix Week 2 rate uses **Week 1 weights** multiplied by each stratum’s **Week 2 completion rate**.
| # | W1 Weight | W2 Rate | Weighted W2 Contribution |
|---:|---:|---:|---:|
| 1 | 0.083333 | 44.0% | 3.6667% |
| 2 | 0.083333 | 48.0% | 4.0000% |
| 3 | 0.083333 | 17.3333% | 1.4433% |
| 4 | 0.083333 | 20.0% | 1.6667% |
| 5 | 0.083333 | 48.0% | 4.0000% |
| 6 | 0.083333 | 52.0% | 4.3333% |
| 7 | 0.083333 | 19.3333% | 1.6111% |
| 8 | 0.083333 | 22.0% | 1.8333% |
| 9 | 0.083333 | 52.0% | 4.3333% |
| 10 | 0.083333 | 56.0% | 4.6667% |
| 11 | 0.083333 | 21.3333% | 1.7778% |
| 12 | 0.083333 | 23.3333% | 1.9444% |
| **Total** | **1.000000** | | **35.2767%** |
Using unrounded Week 2 rates:
```text
Google / skeptical / traceability:
154 / 350 = 0.44
Google / exploring / traceability:
182 / 1,050 = 0.173333…
```
The weighted total is approximately:
```text
35.3%
```
Therefore the comparison is:
| Measure | Week 2 mobile completion |
|---|---:|
| Observed Week 2 rate | **28.1%** |
| Fixed Week 1 mix applied to Week 2 | **35.3%** |
The two are not equal, so composition matters, but the fixed-mix rate is still far below Week 1’s **45.0%**.
## 4. What this comparison does and does not imply
### What it does
- It separates the observed Week 2 mobile rate into two parts:
- the effect of changing stratum mix, and
- the effect of within-stratum rate movement.
- It shows that if Week 2 had preserved the Week 1 stratum mix, the Week 2 mobile completion rate would have been about **35.3%**, not **28.1%**.
- It quantifies the composition effect as roughly:
```text
35.3% − 28.1% ≈ 7.2 percentage points
```
This means part of the observed Week 2 decline is attributable to mobile starts shifting toward lower-completing `exploring` strata and away from higher-completing `skeptical` strata.
### What it does not imply
- It **does not** imply that the remaining decline from Week 1’s 45.0% to the fixed-mix 35.3% is explained by product failure.
- It **does not** show that within-stratum movement has a single cause.
- It **does not** prove that the `exploring` audience performed worse because of the message.
- It **does not** establish causality for either composition change or within-stratum change.
- It **does not** mean the Week 2 aggregate decline is unimportant. Even with fixed Week 1 mix, Week 2 was about **9.7 percentage points below the Week 1 fixed-mix rate**:
```text
45.0% − 35.3% ≈ 9.7 pp
```
The within-stratum drop is still the larger component.
## 5. Arithmetic uncertainty
The table uses counts summed by hand from the supplied CSV structure. I did not execute code. The following should be rechecked in the final reviewed draft:
- exact duplicate row removal and alias normalization;
- exact Week 2 starts for each mobile stratum;
- decimal retention in weighted contributions;
- whether any zero-denominator stratum should be marked unavailable.
The broad conclusion is unlikely to change from small corrections: **both composition and within-stratum movement contributed to the mobile decline; within-stratum movement was the larger contributor.**
## 6. Next evidence needed
To move from composition arithmetic to a defensible explanation, the next evidence is:
1. **Timestamped event-level data** for mobile quiz starts and completions, matched to cohort date and source tag.
2. **Release/QA log** covering the mobile landing page and quiz between Week 1 and Week 2.
3. **Device-level landing-session detail** to verify whether mobile starts became less likely to complete at the same quiz stage.
4. **Audience bucket definitions** to confirm whether `skeptical` and `exploring` shares changed because of targeting/auction changes or because the reporting buckets themselves changed.
Only after those are available can the team test whether the remaining within-stratum mobile drop is a journey problem, a measurement break, or a real behavioral shift.Follow-up prompt
Focused follow-up: quantify the mobile composition question. Show all 12 channel/audience/creative strata with Week 1/2 starts, completions and rates, Week 1 weights, and the weighted Week 2 contribution. State observed Week 2 rate versus fixed Week 1 mix and what each does and does not imply. Recheck any arithmetic you cannot substantiate. Do not infer a product cause from this comparison. Keep this response focused on the calculation and the next evidence needed.Production prompt
Production pass: use ONLY the approved facts and decisions attached below. Format the Founder / CEO Summary, Media Team Analysis and Content & Creative Brief. Do not introduce new claims, calculations, causes, promises, budgets or actions. Then format the approved retrospective Detect → Analyze → Route → Review example including observation window, snapshot, earliest justified decision time, responsible role, cadence, expiry/escalation and human client-facing approval state. This is a static illustration with no live messages or automation. Keep unknown outcomes explicit.
## Attached file: en/review.md
# What changed in review
Fictional GreenLeaf UK, paid-click cohorts September 1–14, 2026; snapshot September 22, 23:59:59 UTC.
This record was prepared with AI-assisted editorial review and independently checked calculations.
It is **not a completed human approval, client review or controlled human-versus-AI experiment**.
The owner’s publication/release decision remains pending. Original model responses are unchanged.
## The decision before the wording
The first draft recommends diagnosis before a creative or budget pivot. That direction is reasonable,
but several numbers and the explanation are wrong. A plausible recommendation does not make its
supporting evidence safe to release.
| Initial draft statement | Checked evidence | Review decision |
|---|---|---|
| Spend was $110,100 in each week | Week 1: $5,462; Week 2: $5,512 | Reject both draft totals and every derived CPA/ROAS based on them |
| New customers were 548 → 482 | Canonical paid-cohort counts: 467 → 357 | Replace; these are attributed first purchases, not incremental subscribers |
| First-purchase revenue was $43,840 → $38,560 | $37,360 → $28,560 | Replace; no causal revenue-loss estimate follows |
| Mobile Week 2 had 2,360 completions | 2,352 / 8,400 = 28% | Replace the count/rate; retain Week 1’s checked 3,780 / 8,400 = 45% |
| Mobile audience/creative shares were stable and composition was essentially absent | Skeptical share of mobile starts: 50% → 25%; exploring: 50% → 75% | Reject the mix explanation |
| Fixed Week 1 mix gives roughly 35.3%–35.5% | Exactly 35⅓% before display rounding | Use 35.33%; state that both mix and within-group changes matter |
| Founder should freeze spend reallocation | The CSV alone does not establish what all commercial/risk decisions should be | Narrow to “do not recommend a budget move from this evidence alone”; retain existing authorized risk processes |
| A diagnostic succeeds if completion returns toward 45% | Returning to a target is not a causal test design | Require verified event definitions, version/cohort evidence and an explicit comparison plan |
All channel, desktop, audience and creative tables in the raw response remain **unapproved**. The
final editorial drafts use the checked metric pack instead. The model invocation was text-only,
without a calculation-execution tool. This is evidence about this run, not every AI analysis product.
## What the focused follow-up did
The second response recognizes the audience-mix shift and gives the correct Week 2 starts per
stratum: skeptical 350, exploring 1,050. It still reports 2,360 total completions, misstates the
Reddit/exploring/value cell as 245 rather than 252, and miscalculates a weighted contribution.
Use the recorded follow-up to inspect iteration; do not silently substitute it for verified math.
## Approved facts for the production draft
| Measure | Week 1 | Week 2 |
|---|---:|---:|
| All-paid media spend | $5,462 | $5,512 |
| All-paid quiz starts | 15,040 | 15,120 |
| All-paid quiz completions | 6,770 | 5,460 |
| Attributed first-purchase customers | 467 | 357 |
| Attributed first-purchase revenue | $37,360 | $28,560 |
| Media CPA | $11.70 | $15.44 |
| Attributed first-order ROAS | 6.84 | 5.18 |
| Mobile quiz starts | 8,400 | 8,400 |
| Mobile quiz completions | 3,780 | 2,352 |
| Mobile completion | 45% | 28% |
| Desktop completion | 45.03% | 46.25% |
Mobile change: −17 percentage points; approximately −37.8% relative. Fixed Week 1 channel ×
audience × creative weights give a Week 2 mobile rate of 35.33%. Under this particular baseline-weight
decomposition, the within-group component is −9.67 points and the remaining composition component
is −7.33 points. Another decomposition order can allocate interaction differently; this is descriptive,
not causal attribution. All 12 comparable mobile strata decline.
## New fictional context and its effect
The context note was prewritten and hashed before the run. It reports a mobile presentation revision
starting September 8 at 09:00 UTC and an unlinked support concern about finding the next-step control.
The release audit, rollout share, ticket counts and links to paid cohorts are missing.
**Verdict: revise the explanation and narrow the next action; confirm the diagnostic-first direction.**
Ask the product owner and media lead to verify the mobile journey and event definitions, request
version-assignment/release evidence, and compare like-for-like groups where evidence permits.
The product owner’s reported availability for a check in the next working day is fictional context,
not evidence that a check happened. A product cause, creative failure or revenue loss remains unknown.
Carry the report of the revision into brief v2 as **reported context pending verification**, never
as a confirmed cause. No automatic pause, spend change or external alert is approved.
## Instructions to the production pass
Preserve the approved table, uncertainty and decision above across three drafts:
- Founder/CEO: trajectory and the diagnostic decision, with at most three actions and role owners.
- Media: period/channel/device detail, the fixed-mix comparison and the investigation work order.
- Content/creative: keep message hypotheses distinct from proof; inspect post-click context before
proposing a message test. Do not present a causal creative winner or recommend media-budget changes.
For the retrospective alert: trigger on settled paid-mobile completion below the stated 45% target.
With end-of-day click cohorts and a seven-day observation rule, the full Week 2 comparison is first
justified September 21 at 23:59:59 UTC; the supplied snapshot is September 22 at 23:59:59 UTC.
Detect → Analyze → Route → Review is a static demonstration. Product owner/media lead own the
diagnostic; a measurement-competent reviewer with release authority owns client-facing approval.
No human approval or external delivery has occurred. Recheck the rule/context next cycle, and
escalate missing definitions or evidence to the marketing owner rather than issuing a firm claim.
## Provenance
Checked against the supplied reconciled CSV, independently reviewed deterministic calculations and
`data/analysis.json`. The raw initial and follow-up responses remain downloadable evidence. Numeric
claims in final editorial drafts must map to the claim ledger. “Approved” here means approved for
this AI-assisted editorial production pass; it does not stand in for the owner’s human release gate.
## Attached file: en/brief-v2.md
# GreenLeaf reporting brief v2 — carry the learning forward
**Status:** AI-assisted editorial draft; human publication/release approval pending. Fictional case,
not client-operating evidence. This is an additive update to [complete brief v1](brief-v1.md), not a
replacement with a shorter prompt. Read both together for the next cycle.
The business goals, KPI definitions, stakeholder responsibilities, attribution rules, target periods
and cadence in v1 remain unchanged. The new knowledge is separated by evidence class below.
## Checked data findings
- Paid-mobile completion is 3,780/8,400 = 45% in Week 1 and 2,352/8,400 = 28% in Week 2.
- The mobile skeptical/exploring starts mix changes from 50%/50% to 25%/75%.
- Using fixed Week 1 channel/audience/creative weights gives 35.33% for Week 2. Both composition and
within-group movement contribute descriptively; all 12 mobile strata decline. No cause is proved.
- The first text-only AI response and its focused follow-up contain arithmetic errors. Reuse the
checked metric pack and claim references, not unverified numbers from those responses.
## Reported fictional context — verification outstanding
The product owner reports a mobile quiz presentation revision beginning September 8 at 09:00 UTC.
The support lead reports difficulty finding the next control on a phone, without dates, counts or
linkage to paid cohorts. No deployment audit, rollout share or version-assignment evidence is supplied.
These reports prioritize an investigation; they are not verified release logs or a causal diagnosis.
## Next-cycle work order
| Owner role | Required evidence/action | Stop condition |
|---|---|---|
| Product owner | Reproduce the mobile flow and retrieve release/version/QA evidence | Missing assignment/logs prevents a version-specific causal claim |
| Media lead | Reconcile event meaning, cohort tagging and comparable device/audience/creative cuts | Inconsistent definitions or immature windows prevent a settled comparison |
| Content lead | Inspect message/post-click expectations after measurement/journey checks | No confirmed message problem means no claim of a winning or failed hook |
| Reviewer with current measurement knowledge and release authority | Check the decision and all stakeholder drafts before client-facing release | Approval absent means no external delivery |
## Hypotheses that must remain hypotheses
Presentation friction, tracking change, targeting/mix changes and other influences may contribute.
The support report does not establish prevalence or abandonment. No quiz-driven revenue loss,
incremental lift, profit, realized LTV, full CAC or monthly-business-target verdict has been established.
Revisit the context, rule and evidence requests at the next reporting cycle; remove stale assumptions
or keep them explicitly unresolved. Neither this update nor its reuse proves improved future output.
## Attached file: data/analysis.json
{
"window": {
"week1Start": "2026-09-01",
"week1End": "2026-09-07",
"week2Start": "2026-09-08",
"week2End": "2026-09-14",
"observationDays": 7,
"snapshot": "2026-09-22T23:59:59Z"
},
"comparisonStatus": "settled",
"periods": {
"week1": {
"spend_cents": 546200,
"impressions": 412560,
"clicks": 21720,
"landing_sessions": 18380,
"quiz_starts": 15040,
"quiz_completions": 6770,
"attributed_new_customers": 467,
"attributed_first_purchase_revenue_cents": 3736000,
"rows": 168,
"settled": true,
"quizCompletionRate": 0.45013297872340424,
"mediaCpaCents": 1169.593147751606,
"attributedFirstOrderRoas": 6.839985353350421,
"ctr": 0.052646887725421754,
"cpcCents": 25.147329650092082,
"cpmCents": 1323.9286406825674
},
"week2": {
"spend_cents": 551200,
"impressions": 414960,
"clicks": 21840,
"landing_sessions": 18480,
"quiz_starts": 15120,
"quiz_completions": 5460,
"attributed_new_customers": 357,
"attributed_first_purchase_revenue_cents": 2856000,
"rows": 168,
"settled": true,
"quizCompletionRate": 0.3611111111111111,
"mediaCpaCents": 1543.9775910364147,
"attributedFirstOrderRoas": 5.181422351233672,
"ctr": 0.05263157894736842,
"cpcCents": 25.238095238095237,
"cpmCents": 1328.3208020050126
}
},
"byChannel": {
"google-search": {
"week1": {
"spend_cents": 173600,
"impressions": 131040,
"clicks": 7280,
"landing_sessions": 6160,
"quiz_starts": 5040,
"quiz_completions": 2142,
"attributed_new_customers": 154,
"attributed_first_purchase_revenue_cents": 1232000,
"rows": 56,
"settled": true,
"quizCompletionRate": 0.425,
"mediaCpaCents": 1127.2727272727273,
"attributedFirstOrderRoas": 7.096774193548387,
"ctr": 0.05555555555555555,
"cpcCents": 23.846153846153847,
"cpmCents": 1324.7863247863247
},
"week2": {
"spend_cents": 174400,
"impressions": 131040,
"clicks": 7280,
"landing_sessions": 6160,
"quiz_starts": 5040,
"quiz_completions": 1722,
"attributed_new_customers": 105,
"attributed_first_purchase_revenue_cents": 840000,
"rows": 56,
"settled": true,
"quizCompletionRate": 0.3416666666666667,
"mediaCpaCents": 1660.952380952381,
"attributedFirstOrderRoas": 4.81651376146789,
"ctr": 0.05555555555555555,
"cpcCents": 23.956043956043956,
"cpmCents": 1330.8913308913309
}
},
"linkedin": {
"week1": {
"spend_cents": 229600,
"impressions": 138320,
"clicks": 7280,
"landing_sessions": 6160,
"quiz_starts": 5040,
"quiz_completions": 2268,
"attributed_new_customers": 154,
"attributed_first_purchase_revenue_cents": 1232000,
"rows": 56,
"settled": true,
"quizCompletionRate": 0.45,
"mediaCpaCents": 1490.909090909091,
"attributedFirstOrderRoas": 5.365853658536586,
"ctr": 0.05263157894736842,
"cpcCents": 31.53846153846154,
"cpmCents": 1659.919028340081
},
"week2": {
"spend_cents": 230400,
"impressions": 138320,
"clicks": 7280,
"landing_sessions": 6160,
"quiz_starts": 5040,
"quiz_completions": 1820,
"attributed_new_customers": 119,
"attributed_first_purchase_revenue_cents": 952000,
"rows": 56,
"settled": true,
"quizCompletionRate": 0.3611111111111111,
"mediaCpaCents": 1936.1344537815125,
"attributedFirstOrderRoas": 4.131944444444445,
"ctr": 0.05263157894736842,
"cpcCents": 31.64835164835165,
"cpmCents": 1665.7027183342973
}
},
"reddit-paid": {
"week1": {
"spend_cents": 143000,
"impressions": 143200,
"clicks": 7160,
"landing_sessions": 6060,
"quiz_starts": 4960,
"quiz_completions": 2360,
"attributed_new_customers": 159,
"attributed_first_purchase_revenue_cents": 1272000,
"rows": 56,
"settled": true,
"quizCompletionRate": 0.47580645161290325,
"mediaCpaCents": 899.37106918239,
"attributedFirstOrderRoas": 8.895104895104895,
"ctr": 0.05,
"cpcCents": 19.972067039106147,
"cpmCents": 998.6033519553073
},
"week2": {
"spend_cents": 146400,
"impressions": 145600,
"clicks": 7280,
"landing_sessions": 6160,
"quiz_starts": 5040,
"quiz_completions": 1918,
"attributed_new_customers": 133,
"attributed_first_purchase_revenue_cents": 1064000,
"rows": 56,
"settled": true,
"quizCompletionRate": 0.38055555555555554,
"mediaCpaCents": 1100.751879699248,
"attributedFirstOrderRoas": 7.26775956284153,
"ctr": 0.05,
"cpcCents": 20.10989010989011,
"cpmCents": 1005.4945054945055
}
}
},
"byCreative": {
"traceability": {
"week1": {
"spend_cents": 274400,
"impressions": 207480,
"clicks": 10920,
"landing_sessions": 9240,
"quiz_starts": 7560,
"quiz_completions": 3276,
"attributed_new_customers": 231,
"attributed_first_purchase_revenue_cents": 1848000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.43333333333333335,
"mediaCpaCents": 1187.878787878788,
"attributedFirstOrderRoas": 6.73469387755102,
"ctr": 0.05263157894736842,
"cpcCents": 25.128205128205128,
"cpmCents": 1322.5371120107961
},
"week2": {
"spend_cents": 275600,
"impressions": 207480,
"clicks": 10920,
"landing_sessions": 9240,
"quiz_starts": 7560,
"quiz_completions": 2625,
"attributed_new_customers": 161,
"attributed_first_purchase_revenue_cents": 1288000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.3472222222222222,
"mediaCpaCents": 1711.8012422360248,
"attributedFirstOrderRoas": 4.67343976777939,
"ctr": 0.05263157894736842,
"cpcCents": 25.238095238095237,
"cpmCents": 1328.3208020050126
}
},
"value": {
"week1": {
"spend_cents": 271800,
"impressions": 205080,
"clicks": 10800,
"landing_sessions": 9140,
"quiz_starts": 7480,
"quiz_completions": 3494,
"attributed_new_customers": 236,
"attributed_first_purchase_revenue_cents": 1888000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.46711229946524063,
"mediaCpaCents": 1151.6949152542372,
"attributedFirstOrderRoas": 6.946284032376748,
"ctr": 0.0526623756582797,
"cpcCents": 25.166666666666668,
"cpmCents": 1325.3364540667058
},
"week2": {
"spend_cents": 275600,
"impressions": 207480,
"clicks": 10920,
"landing_sessions": 9240,
"quiz_starts": 7560,
"quiz_completions": 2835,
"attributed_new_customers": 196,
"attributed_first_purchase_revenue_cents": 1568000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.375,
"mediaCpaCents": 1406.1224489795918,
"attributedFirstOrderRoas": 5.689404934687953,
"ctr": 0.05263157894736842,
"cpcCents": 25.238095238095237,
"cpmCents": 1328.3208020050126
}
}
},
"byAudience": {
"skeptical": {
"week1": {
"spend_cents": 274400,
"impressions": 207480,
"clicks": 10920,
"landing_sessions": 9240,
"quiz_starts": 7560,
"quiz_completions": 4200,
"attributed_new_customers": 301,
"attributed_first_purchase_revenue_cents": 2408000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.5555555555555556,
"mediaCpaCents": 911.6279069767442,
"attributedFirstOrderRoas": 8.775510204081632,
"ctr": 0.05263157894736842,
"cpcCents": 25.128205128205128,
"cpmCents": 1322.5371120107961
},
"week2": {
"spend_cents": 275600,
"impressions": 183540,
"clicks": 9660,
"landing_sessions": 7980,
"quiz_starts": 6300,
"quiz_completions": 3108,
"attributed_new_customers": 203,
"attributed_first_purchase_revenue_cents": 1624000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.49333333333333335,
"mediaCpaCents": 1357.6354679802955,
"attributedFirstOrderRoas": 5.8925979680696665,
"ctr": 0.05263157894736842,
"cpcCents": 28.530020703933747,
"cpmCents": 1501.5800370491445
}
},
"exploring": {
"week1": {
"spend_cents": 271800,
"impressions": 205080,
"clicks": 10800,
"landing_sessions": 9140,
"quiz_starts": 7480,
"quiz_completions": 2570,
"attributed_new_customers": 166,
"attributed_first_purchase_revenue_cents": 1328000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.34358288770053474,
"mediaCpaCents": 1637.3493975903614,
"attributedFirstOrderRoas": 4.885945548197204,
"ctr": 0.0526623756582797,
"cpcCents": 25.166666666666668,
"cpmCents": 1325.3364540667058
},
"week2": {
"spend_cents": 275600,
"impressions": 231420,
"clicks": 12180,
"landing_sessions": 10500,
"quiz_starts": 8820,
"quiz_completions": 2352,
"attributed_new_customers": 154,
"attributed_first_purchase_revenue_cents": 1232000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.26666666666666666,
"mediaCpaCents": 1789.6103896103896,
"attributedFirstOrderRoas": 4.470246734397678,
"ctr": 0.05263157894736842,
"cpcCents": 22.627257799671593,
"cpmCents": 1190.9083052458734
}
}
},
"byDevice": {
"mobile": {
"week1": {
"spend_cents": 274400,
"impressions": 223440,
"clicks": 11760,
"landing_sessions": 10080,
"quiz_starts": 8400,
"quiz_completions": 3780,
"attributed_new_customers": 259,
"attributed_first_purchase_revenue_cents": 2072000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.45,
"mediaCpaCents": 1059.4594594594594,
"attributedFirstOrderRoas": 7.551020408163265,
"ctr": 0.05263157894736842,
"cpcCents": 23.333333333333332,
"cpmCents": 1228.0701754385964
},
"week2": {
"spend_cents": 275600,
"impressions": 223440,
"clicks": 11760,
"landing_sessions": 10080,
"quiz_starts": 8400,
"quiz_completions": 2352,
"attributed_new_customers": 147,
"attributed_first_purchase_revenue_cents": 1176000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.28,
"mediaCpaCents": 1874.8299319727892,
"attributedFirstOrderRoas": 4.267053701015965,
"ctr": 0.05263157894736842,
"cpcCents": 23.435374149659864,
"cpmCents": 1233.44074471894
}
},
"desktop": {
"week1": {
"spend_cents": 271800,
"impressions": 189120,
"clicks": 9960,
"landing_sessions": 8300,
"quiz_starts": 6640,
"quiz_completions": 2990,
"attributed_new_customers": 208,
"attributed_first_purchase_revenue_cents": 1664000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.4503012048192771,
"mediaCpaCents": 1306.7307692307693,
"attributedFirstOrderRoas": 6.12214863870493,
"ctr": 0.05266497461928934,
"cpcCents": 27.289156626506024,
"cpmCents": 1437.1827411167512
},
"week2": {
"spend_cents": 275600,
"impressions": 191520,
"clicks": 10080,
"landing_sessions": 8400,
"quiz_starts": 6720,
"quiz_completions": 3108,
"attributed_new_customers": 210,
"attributed_first_purchase_revenue_cents": 1680000,
"rows": 84,
"settled": true,
"quizCompletionRate": 0.4625,
"mediaCpaCents": 1312.3809523809523,
"attributedFirstOrderRoas": 6.095791001451379,
"ctr": 0.05263157894736842,
"cpcCents": 27.341269841269842,
"cpmCents": 1439.014202172097
}
}
},
"mobile": {
"week1Rate": 0.45,
"week2Rate": 0.28,
"week1Starts": 8400,
"week2Starts": 8400,
"week1Completions": 3780,
"week2Completions": 2352,
"percentagePointChange": -17,
"relativeChange": -0.3777777777777777,
"comparableStrata": true,
"fixedMixWeek2Rate": 0.3533333333333334,
"strata": [
{
"key": "google-search/exploring/traceability",
"week1Starts": 700,
"week2Starts": 1050,
"week1Completions": 175,
"week2Completions": 182,
"week1Rate": 0.25,
"week2Rate": 0.17333333333333334,
"baselineWeight": 0.08333333333333333
},
{
"key": "google-search/exploring/value",
"week1Starts": 700,
"week2Starts": 1050,
"week1Completions": 203,
"week2Completions": 210,
"week1Rate": 0.29,
"week2Rate": 0.2,
"baselineWeight": 0.08333333333333333
},
{
"key": "google-search/skeptical/traceability",
"week1Starts": 700,
"week2Starts": 350,
"week1Completions": 378,
"week2Completions": 154,
"week1Rate": 0.54,
"week2Rate": 0.44,
"baselineWeight": 0.08333333333333333
},
{
"key": "google-search/skeptical/value",
"week1Starts": 700,
"week2Starts": 350,
"week1Completions": 406,
"week2Completions": 168,
"week1Rate": 0.58,
"week2Rate": 0.48,
"baselineWeight": 0.08333333333333333
},
{
"key": "linkedin/exploring/traceability",
"week1Starts": 700,
"week2Starts": 1050,
"week1Completions": 196,
"week2Completions": 203,
"week1Rate": 0.28,
"week2Rate": 0.19333333333333333,
"baselineWeight": 0.08333333333333333
},
{
"key": "linkedin/exploring/value",
"week1Starts": 700,
"week2Starts": 1050,
"week1Completions": 224,
"week2Completions": 231,
"week1Rate": 0.32,
"week2Rate": 0.22,
"baselineWeight": 0.08333333333333333
},
{
"key": "linkedin/skeptical/traceability",
"week1Starts": 700,
"week2Starts": 350,
"week1Completions": 406,
"week2Completions": 168,
"week1Rate": 0.58,
"week2Rate": 0.48,
"baselineWeight": 0.08333333333333333
},
{
"key": "linkedin/skeptical/value",
"week1Starts": 700,
"week2Starts": 350,
"week1Completions": 434,
"week2Completions": 182,
"week1Rate": 0.62,
"week2Rate": 0.52,
"baselineWeight": 0.08333333333333333
},
{
"key": "reddit-paid/exploring/traceability",
"week1Starts": 700,
"week2Starts": 1050,
"week1Completions": 217,
"week2Completions": 224,
"week1Rate": 0.31,
"week2Rate": 0.21333333333333335,
"baselineWeight": 0.08333333333333333
},
{
"key": "reddit-paid/exploring/value",
"week1Starts": 700,
"week2Starts": 1050,
"week1Completions": 245,
"week2Completions": 252,
"week1Rate": 0.35,
"week2Rate": 0.24,
"baselineWeight": 0.08333333333333333
},
{
"key": "reddit-paid/skeptical/traceability",
"week1Starts": 700,
"week2Starts": 350,
"week1Completions": 434,
"week2Completions": 182,
"week1Rate": 0.62,
"week2Rate": 0.52,
"baselineWeight": 0.08333333333333333
},
{
"key": "reddit-paid/skeptical/value",
"week1Starts": 700,
"week2Starts": 350,
"week1Completions": 462,
"week2Completions": 196,
"week1Rate": 0.66,
"week2Rate": 0.56,
"baselineWeight": 0.08333333333333333
}
]
},
"unavailable": [
"true-cac",
"incrementality",
"awareness",
"realized-ltv",
"profitability"
]
}
Original production response
# GreenLeaf Wellness UK — Production Reporting Draft
**Fictional teaching example.** Paid-click cohorts 2026-09-01 through 2026-09-14; snapshot 2026-09-22T23:59:59Z.
**Basis:** approved facts in `review.md`, `brief-v2.md`, and `analysis.json`.
**Release state:** AI-assisted editorial production pass only. No human approval, client review, or external delivery has occurred.
---
# 1. Founder / CEO Summary
## WHAT happened
Mobile quiz completion fell from **45% in Week 1** to **28% in Week 2**:
| Mobile metric | Week 1 | Week 2 |
|---|---:|---:|
| Quiz starts | 8,400 | 8,400 |
| Quiz completions | 3,780 | 2,352 |
| Completion rate | 45% | 28% |
That is −17 percentage points, approximately −37.8% relative.
The decline is not a mirage of device totals. It sits alongside a mobile audience-mix change: skeptical share of mobile starts fell from 50% to 25%, while exploring share rose from 50% to 75%. When Week 2 is reweighted using Week 1 channel × audience × creative weights, the mobile completion rate is **35.33%** rather than 28%. Both composition and within-group movement therefore contribute descriptively. All 12 mobile strata declined.
At the all-paid period level:
| Measure | Week 1 | Week 2 |
|---|---:|---:|
| All-paid media spend | $5,462 | $5,512 |
| All-paid quiz starts | 15,040 | 15,120 |
| All-paid quiz completions | 6,770 | 5,460 |
| Attributed first-purchase customers | 467 | 357 |
| Attributed first-purchase revenue | $37,360 | $28,560 |
| Media CPA | $11.70 | $15.44 |
| Attributed first-order ROAS | 6.84 | 5.18 |
Media CPA increased and attributed first-order ROAS declined between the two mature paid-cohort periods.
## WHY it matters
The quiz is the main early signal for UK consideration. A 17-point mobile completion decline means the launch is losing progress at the point where interested buyers should become qualified first-purchase candidates. The decline is broad across mobile strata, so the immediate priority is to establish whether the mobile post-click journey or measurement definition changed.
The financial picture is directionally weaker, but it remains **paid-cohort attributed first-purchase revenue only**. No profit, full CAC, incrementality, or all-channel monthly revenue conclusion is available.
## WHAT’S NEXT — at most three actions
1. **Do not recommend a budget move from this evidence alone.**
Owner: founder / marketing owner. Existing authorized risk and commercial processes remain the approval route.
2. **Verify the mobile journey and event definitions.**
Owner: product owner + media lead. Apply the diagnostic work order in Section 2.5. No version-specific causal claim is approved before release/version/QA evidence is supplied.
3. **Hold the diagnostic decision until role owners complete their checks.**
Owner: founder / marketing owner, with a measurement-competent reviewer holding client-facing release authority. No external or client-facing delivery has occurred.
**Unavailable and not to be inferred:** true CAC, incrementality, awareness, realized LTV, profitability, full-month revenue, and any causal revenue-loss estimate.
---
# 2. Media Team Analysis
## 2.1 Comparison status
| Item | Status |
|---|---|
| Week 1 dates | 2026-09-01 through 2026-09-07 |
| Week 2 dates | 2026-09-08 through 2026-09-14 |
| Observation window | 7 days post-click for each cohort |
| Snapshot | 2026-09-22T23:59:59Z |
| Comparison status | Settled |
## 2.2 Period totals
### All paid
| Measure | Week 1 | Week 2 |
|---|---:|---:|
| Spend | $5,462 | $5,512 |
| Impressions | 412,560 | 414,960 |
| Clicks | 21,720 | 21,840 |
| Landing sessions | 18,380 | 18,480 |
| Quiz starts | 15,040 | 15,120 |
| Quiz completions | 6,770 | 5,460 |
| Completion rate | 45.01% | 36.11% |
| Attributed first-purchase customers | 467 | 357 |
| Attributed first-purchase revenue | $37,360 | $28,560 |
| Media CPA | $11.70 | $15.44 |
| Attributed first-order ROAS | 6.84 | 5.18 |
### Device
| Device | Week | Starts | Completions | Completion rate |
|---|---|---:|---:|---:|
| Mobile | W1 | 8,400 | 3,780 | 45.00% |
| Mobile | W2 | 8,400 | 2,352 | 28.00% |
| Desktop | W1 | 6,640 | 2,990 | 45.03% |
| Desktop | W2 | 6,720 | 3,108 | 46.25% |
Mobile starts stayed flat at 8,400. Mobile completion fell; desktop completion moved slightly upward.
## 2.3 Mobile audience-mix change
| Audience | Mobile starts W1 | Mobile starts W2 | Share W1 | Share W2 |
|---|---:|---:|---:|---:|
| Skeptical | 4,200 | 2,100 | 50% | 25% |
| Exploring | 4,200 | 6,300 | 50% | 75% |
This is a material composition change. The correct Week 2 mobile strata are:
- skeptical strata: 350 mobile starts each;
- exploring strata: 1,050 mobile starts each.
## 2.4 Fixed Week 1 mix calculation
Using Week 1 channel × audience × creative quiz-start shares as fixed weights:
| Calculation | Result |
|---|---:|
| Week 1 observed mobile completion | 45.00% |
| Week 2 observed mobile completion | 28.00% |
| Week 2 fixed-mix mobile completion | **35.33%** |
Under this particular baseline-weight decomposition:
| Component | Contribution |
|---|---:|
| Within-group change | −9.67 points |
| Remaining composition component | −7.33 points |
These are descriptive decomposition components, not causal attribution. A different decomposition order may allocate interaction differently.
Because all 12 comparable mobile strata declined, the fixed-mix result cannot be interpreted as a simple composition-only story. The mobile decline has both within-group and composition elements.
## 2.5 Investigation work order
| Owner role | Required evidence/action | Stop condition |
|---|---|---|
| Product owner | Reproduce the mobile flow and retrieve release/version/QA evidence | Missing assignment/logs prevents a version-specific causal claim |
| Media lead | Reconcile event meaning, cohort tagging, and comparable device/audience/creative cuts | Inconsistent definitions or immature windows prevent a settled comparison |
| Reviewer with current measurement knowledge and release authority | Check the decision and all stakeholder drafts before client-facing release | Approval absent means no external delivery |
**Reported fictional context pending verification, not confirmed cause:**
- Product owner reports a mobile quiz presentation revision beginning September 8 at 09:00 UTC.
- Support lead reports an unlinked concern about finding the next-step control.
No deployment audit, rollout share, ticket counts, or linkage to paid cohorts has been supplied. These reports prioritize an investigation; they are not evidence of causality.
---
# 3. Content & Creative Brief
## Message/post-click hypotheses
The broad mobile decline could reflect:
1. **Presentation friction.** The mobile quiz presentation may have changed after Week 1.
2. **Measurement/tracking change.** Event definitions or source tagging may differ between Week 1 and Week 2.
3. **Targeting/mix changes.** The mobile audience mix shifted from 50% skeptical to 75% exploring.
These remain hypotheses.
No creative or message winner is claimed. The decline appears across both `traceability` and `value` mobile strata, and across all comparable mobile strata.
## Focused next step
Inspect the post-click context before proposing a message test.
| Component | Detail |
|---|---|
| Owner | Content lead, product owner |
| First step | Review post-click journey evidence after the media/product diagnostic begins |
| Required before message testing | Verified mobile event definitions, version-assignment evidence, and comparable cohort context |
| Stop condition | If measurement or journey remains unverified, do not claim a failed or winning hook |
| Intent | Keep message hypotheses distinct from proof |
---
# 4. Retrospective Alert Example
## Trigger
Settled paid-mobile quiz completion below the stated 45% target.
## Window
- Observation window: 7 days after each end-of-day paid-click cohort.
- Week 2 comparison dates: 2026-09-08 through 2026-09-14.
- Snapshot: **2026-09-22T23:59:59Z**.
## Earliest justified detection time
With end-of-day click cohorts and a seven-day observation rule, the full Week 2 paid-mobile comparison is first justified at:
```text
2026-09-21T23:59:59Z
```
The supplied snapshot is:
```text
2026-09-22T23:59:59Z
```
## Detect → Analyze → Route → Review
| Stage | Action | Responsible role |
|---|---|---|
| Detect | Settled paid-mobile completion is below 45% | Measurement/performance lead |
| Analyze | Compare both periods, device/audience/creative cuts, fixed-mix check | Media lead |
| Route | Assign mobile journey and event-definition verification | Product owner + media lead |
| Review | Check evidence and approve any client-facing statement | Measurement-competent reviewer with release authority |
## Cadence
This is a static illustration for the next reporting cycle. No live messages, automated spend changes, or external alerts are active.
## Expiry / escalation
- Recheck the rule, context, and evidence requests at the next reporting cycle.
- Escalate missing definitions or evidence to the marketing owner rather than issuing a firm claim.
## Human client-facing approval state
```text
No human approval has occurred.
No client-facing release has occurred.
```
---
# 5. Limitations
## Unavailable outcomes
The following are explicitly unavailable:
- True CAC
- Incrementality
- Awareness
- Realized LTV
- Profitability
- Full-month all-channel revenue
- Causal revenue-loss estimate
## Static demonstration
This retrospective alert is a static teaching demonstration. It does not imply running automation, real-time detection, or externally visible alerting.
## Fictional case
This is a constructed example. It is not real client-operating evidence and does not represent real customers, accounts, approvals, or campaign results.
---
# 6. Use of approved facts only
This production draft uses only the approved tables, calculations, and decisions from `review.md`, `brief-v2.md`, and `analysis.json`. The metrics above are consistent across all three deliverables:
- Mobile completion: **45% → 28%**
- Fixed-mix Week 2 mobile: **35.33%**
- All-paid spend: **$5,462 → $5,512**
- All-paid completions: **6,770 → 5,460**
- Attributed first-purchase customers: **467 → 357**
- Attributed first-purchase revenue: **$37,360 → $28,560**
- Media CPA: **$11.70 → $15.44**
- Attributed first-order ROAS: **6.84 → 5.18**
- Desktop completion: **45.03% → 46.25%**Run provenance and limits
{
"version": 1,
"status": "editorial-review-human-release-pending",
"fictionalCampaign": true,
"actualModelInvocations": true,
"provider": "deepseek",
"requestedModel": "deepseek-v4-pro",
"returnedModel": "deepseek-v4-pro",
"sourceCommit": "40fbb5bbf84f77e38205bfc9c88a34ab3100a289",
"initialAttachedFiles": [
"en/brief-v1.md",
"en/dictionary.md",
"data/reconciled.csv",
"data/campaigns.csv",
"data/creatives.csv"
],
"precommittedContext": {
"path": "context-note.md",
"sha256": "ae57dd174c3b596afd097ff89a3d665ab27b809767713a98ef9d6e03e9646dec",
"includedInInitialPrompt": false,
"explanation": "The legacy inputHashes field also records the precommitted context hash. The actual request messages exclude its contents; it is provenance, not an initial attachment."
},
"evidence": {
"prompt.md": "b7526a2e10958f90e1527b57852e4a946369dc571be041584a3de8578ed2dc2b",
"response.md": "af3417c37a8e507c504ae99192b805af300f5c4d4f1ba18f3bceef070d504677",
"draft-attempt-01.json": "58ffc1a277245483703d2e239f98cf6e855f8ef1acc102bac6e9092188a7f92b",
"draft-result.json": "5c38314c6b12a5cb891a6fc944239364fe03528569cd6581abe5c377e3cf257e",
"draft-source.json": "268a719071e323171ad784011db92006a4aac952c6fd3bb5f38d5bb7044f1f68",
"analysis-prompt.md": "92114f776445befa6c3ad94301b3387fd1671c32acbeb8159466511691bd3fb0",
"analysis-followup.md": "cbc408a78a97010c8d27790257d0311f813bc2a666ecc0d17341c51038a274fb",
"analysis-attempt-01.json": "fdcd45729f5a633fbd8b2c307b597e6bd7e5f74b5315931164626d910f0127f9",
"analysis-result.json": "1e93962f23469e14b4ecbc38fc585256f71478d03e6c645761d179ffd9244945",
"analysis-source.json": "268a719071e323171ad784011db92006a4aac952c6fd3bb5f38d5bb7044f1f68",
"context-note.md": "ae57dd174c3b596afd097ff89a3d665ab27b809767713a98ef9d6e03e9646dec",
"production-prompt.md": "130af408bec4c0d6394e89df364f1604bf2c4c6c9c41a2fdcd5a8aaa284fad09",
"production-response.md": "9a59340818e12170f94589818b7f99dbfd00bf7550c49230ceeef59627c3cea2",
"production-attempt-01.json": "e7b921293b530f126e8cfeef9e0e894f905274c131b3c68b84b2b756f4603b9f",
"production-result.json": "777bcf18204a9d9ceab4bd539f258cd675c24dcf3146e87666898a70c7f4e2d9",
"production-source.json": "268a719071e323171ad784011db92006a4aac952c6fd3bb5f38d5bb7044f1f68",
"production-review-input.md": "f7c7472ec2c70eaa5a7c28356f464a80be6adb07c1ba51a6b97ebd1ef0796a81"
},
"limits": [
"One text-only initial invocation, one focused follow-up and one formatting pass; no code-execution tool.",
"The authored mobile decline is not an AI-discovered real campaign anomaly.",
"Calculations and editorial review were AI-assisted; no completed human or client approval is claimed.",
"No time saving, profitability, causal lift or operating-model outcome was measured."
],
"editorialFinalization": "Separate checked editorial reports, not verbatim production-response copies; human release remains pending.",
"productionInputSnapshot": "production-review-input.md preserves the exact review supplied to formatting; en/review.md later gained the production-pass review."
}
Pre-recorded fictional context note
# Supplemental context — prewritten fictional product/support follow-up
**Authored:** 2026-10-29, before the initial AI run. **Use:** introduce only after the initial v1 response, during human review. Record this file's SHA-256 before the run. This is authored teaching context, not a real interview, ticket export, deployment log or finding from a live client. Nothing here describes or predicts the model's response.
## What v1 knew
The brief contained the supplied course business, research hypotheses, measurement definitions and activation handoff. It explicitly marked product release history, mobile QA and support feedback as unknown and assigned follow-up to their owners. No product defect, founder sensitivity or known release was withheld from those supplied v1 inputs. The fictional reporting team had not yet received this follow-up. The real author writes it in advance so later context cannot be tailored to manufacture an AI failure.
## Newly supplied in the fictional review sequence
At the product/support follow-up after the v1 draft, the product owner reports that a mobile-only quiz presentation revision began rolling out on **September 8, 2026, at 09:00 UTC**. It added an expandable ingredient-information panel and changed the label beside the next-step control. The desktop presentation was unchanged. The owner has not supplied the deployment audit, rollout share, version-assignment log or device/browser QA results. This is a reported release description within the fiction, not independently verified instrumentation evidence.
The support lead reports that recent messages included difficulty finding the next-step control on a phone. The supplied note contains no ticket-level dates, counts, customer identifiers or linkage to paid cohorts, quiz events, creative or release version. It does not establish complaint prevalence, whether users abandoned, whether the comments began before the revision or whether any loss was incremental.
The product owner can assign a mobile reproduction and event-verification check in the next working day, with the media lead supplying cohort/funnel definitions. A proposed check should compare affected and unaffected presentation versions where assignment can be reconstructed, verify that starts/completions mean the same thing across versions, and inspect device/browser combinations. If assignment or event evidence is missing, record that limit instead of claiming a controlled test.
## What this can change — and what remains unresolved
This context can make product/measurement verification a more specific or timely next step than an immediate message rewrite or channel move. It can also confirm a sensible existing diagnostic recommendation. The reviewer must decide from the actual draft and checked data; no changed recommendation or reduced confidence is required.
Timing, device scope and a reported usability concern make the revision an **investigable hypothesis**. They do not establish that it caused the aggregate change. Audience/channel/creative mix, within-group movement, event-definition changes, other releases and unrelated influences still require scrutiny. The fixed-mix comparison remains necessary and cannot be replaced by this story. No revenue-loss estimate, causal attribution, automatic pause or exact budget reallocation follows from this note.
Carry into brief v2 as **reported fictional context pending verification**, naming product owner and media lead, requested logs/checks and the unresolved causal question. Record independently verified counts as data findings, reported context as reported context and review actions as decisions. Keep the actual initial response unedited. This is an example of learning after a brief, not a controlled comparison of human and AI judgment.
Use the evidence pack
Public example files. Paid course materials stay separate.
- Reporting brief v1 ↓
- Data dictionary ↓
- Editorial review record ↓
- Updated reporting brief v2 ↓
- What changed in the brief ↓
- Production and routing record ↓
- Original synthetic CSV ↓
- Reconciled CSV ↓
- Checked calculations ↓
- Data issue log ↓
- Campaign lookup ↓
- Creative lookup ↓
- Course facts and reconstruction source map ↓
- Initial submitted prompt ↓