Price the judgement and the outcome, not the hours. AI breaks the link between time and cost, so an hourly model punishes you for getting faster: the better your workflow, the fewer hours you bill and the less you earn for the same result. The AI-native fix is to price what the client is actually buying — the decision, the standard, the result — and treat production as an input cost you control.
I have priced agency work from both sides, as the seller and as the client, and this is the change that trips people up most. The instinct is to keep the rate card and just deliver faster. That is the one move guaranteed to shrink your revenue while your quality stays the same.
The problem with hourly pricing
Hourly pricing made sense when time tracked cost. If a plan took forty hours, it cost roughly forty hours of salary plus overhead, and the rate was that cost with a margin. The model was honest because the input and the output were linked: more work, more hours, more cost. Clients could compare agencies on rate and roughly on effort.
AI severs that link. The same plan might take a fraction of the time, at a fraction of the cost, with the same or better quality. Under the hourly model, that improvement flows entirely to the client as a smaller invoice, and the agency captures none of the value it created. Worse, it gives the agency an incentive to work slowly — which is exactly the behaviour a client does not want to pay for. Hourly pricing is now a structural mismatch, not a stylistic choice.
What you are actually selling
Strip an AI-native engagement back and the client is not buying hours. They are buying a decision they trust, a standard that holds, and a result they could not reliably produce themselves. The model and the workflow are how you deliver that, not what you sell. Once you see it that way, the pricing question becomes: what is the value of the decision, and how do we share it?
This reframe matters because it changes what "scope" means. Under hours, scope is effort. Under outcomes, scope is the result and the constraints around it. A client who wants a campaign that hits a specific target is buying a different thing from a client who wants a fixed volume of deliverables, and the price should reflect which one they need.
Three pricing models
Retainer for judgement. The client pays for access to senior judgement and a standard, on an ongoing basis. This suits work where the value is continuous — strategy, review, the standing call on what to do next — and where the volume is hard to predict. The retainer is not a bucket of hours; it is a subscription to the decision layer, with a defined level of access and accountability.
Project or outcome pricing. The client pays a fixed price for a defined result. This suits work with a clear end state: a plan, a launch, a relaunch, a measurement design. You absorb the production cost risk, which is now low because AI makes production cheap, and you keep the difference if you are efficient. The discipline here is scoping the outcome precisely, so "done" is not a negotiation.
Performance or shared upside. The client pays partly against the result. This suits work where the outcome is measurable and attributable, and where you have enough control to be accountable for it. It is the highest-risk model and the one most prone to argument, so it works only when the measurement is agreed before the work starts. Use it where you genuinely believe in the outcome, not to win a pitch on price.
Most AI-native agencies run a mix: a retainer for the judgement layer, project pricing for defined deliverables, and occasional performance deals where the fit is real.
How to set a price without hours
You can still use cost as a floor, even when it is no longer the basis. Estimate the fully loaded cost of the senior time and the workflow you will commit, add the margin you need, and treat that as the floor below which the work is not worth taking. Then price against value — what the outcome is worth to the client — and the floor stops being the price and becomes the line you will not cross.
The practical version: price the decision, cost the delivery, and never let the second determine the first. If the value-based price is below your cost floor, the work is a bad fit, not a discount opportunity. If it is far above, you have found where the judgement is genuinely worth paying for — which is the whole point of the model.
What changes in the pitch
Outcome pricing changes how you sell, not just how you bill. Instead of presenting a team and a rate card, you present the decision and the result, and you explain why senior judgement plus a workflow is a better bet than a larger team. That is a harder pitch to a procurement-led buyer trained to compare hourly rates, and an easier one to a client who has been burned by slow, expensive work. Know which buyer you are in front of before you lead with the model.
The other pitch change is honesty about what the price includes. Retainer pricing has to name the level of access and accountability, or the client will treat it as an unlimited bucket. Project pricing has to name the outcome and the exclusions. The price is only defensible when the buyer can see exactly what they are buying and what they are not.
The scope problem
Outcome pricing fails most often on scope, not on numbers. A vague outcome invites the client to expand it, and each expansion is work you did not price. The fix is to define the deliverable, the constraints, and what is explicitly out of scope, in the same breath as the price. "A measurement design for the campaign" is scoped; "help us with measurement" is not.
This is the same discipline the agency-design framework applies to services: know what the client is buying beyond the output. The price is defensible only if the outcome is defined. A fixed price with fuzzy scope is not outcome pricing; it is an open-ended favour.
Where the saving goes
The value AI creates — the gap between the old cost of delivery and the new one — has to go somewhere. You can pass it all to the client as cheaper work, keep it all as margin, or share it. All three are defensible, but choose deliberately. Passing it all turns you into a commodity and invites a race to the bottom. Keeping it all works until a client notices and asks why the price did not move. Sharing it is usually the durable answer: the client gets a better deal, you keep a better margin, and the relationship survives the next procurement review.
The mistake is to share it without redesigning first. If you discount before you have rebuilt the delivery, you are giving away margin you never captured. Redesign the workflow, confirm the new cost base, then decide how to split the gain. The cheaper-agency question covers that split in more detail.
What to do on Monday
Take your current rate card and mark every service as either judgement-led or production-led. Judgement-led work moves toward a retainer or an outcome price; production-led work is where you are most exposed to the hourly trap and should be repriced or repositioned first. Then build the cost floor for each service so you know the line you will not cross.
Finally, write down the outcome for your next proposal in one sentence the client could repeat. If you cannot, the scope is not ready to price, and no pricing model will save it.
FAQ
Should we abandon hourly billing entirely?
Not always, but treat it as a fallback for genuinely uncertain work, not the default. Hours remain useful when neither side can define the outcome in advance. The goal is to make outcome or retainer pricing the norm and hourly the exception.
How do we price when the client insists on hourly?
Price the senior time, not the production. If the client will only buy hours, bill the judgement at a rate that reflects its scarcity and let the model's speed be invisible efficiency you do not itemise. Do not pass the production saving through as a lower rate.
What if the client asks for a discount immediately?
A discount on an outcome price is a scope conversation, not a price conversation. Ask what part of the outcome they want to drop, or what constraint they want to relax. Discounting delivery without changing scope just moves the work below your floor.
Can a small agency use performance pricing?
Only where the outcome is measurable, attributable, and within your control. A small shop with a clear read on its work can use it well; one without that read is taking on risk it cannot manage. Start with retainers and projects, and earn the right to performance deals.
The short version
Price the judgement and the outcome, not the hours, because AI has broken the link between time and cost. Use retainers for the decision layer, project pricing for defined outcomes, and performance deals only where the measurement is honest. Cost the delivery, price the value, and never let the second be set by the first. The Agency-Model guide covers the rest, and the for-agencies track works through pricing alongside structure.
Pricing is where an agency's strategy becomes visible. If you have moved off hours, I would like to hear what the client said when you did.
Cheers, Chandler