A quick disclosure before I start: this post is largely researched and written with AI. I used AI chat to test the argument, push back on my assumptions, and assemble the analysis. I am publishing it that way deliberately, because the point of the post is not only the Publicis-LiveRamp deal. It is also a small demonstration of how industry analysis itself is changing when AI can help research, challenge, and write very quickly.
Over the weekend, I saw the news that Publicis has agreed to acquire LiveRamp for about $2.2B.
One line in the announcement made the intended framing very clear: Arthur Sadoun connected Epsilon to helping clients "take back control of their data from the walled gardens." I think that is a useful starting point, but not the whole story.
My first reaction was probably not the reaction the press release wanted me to have.
I did not think: wow, this changes everything.
I thought: does this really matter when advertising is already dominated by walled gardens?
Because that is the honest starting point, right? A large part of digital advertising already happens inside massive closed ecosystems: large search, video, social, retail, marketplace, super-app, and commerce networks.
When I say walled gardens, I mean closed ecosystems that own the inventory, identity, auction, optimization logic, and measurement interface.
They have login data, purchase data, exposure data, bidding systems, creative tools, conversion APIs, and their own measurement dashboards.
So why should anyone care that an agency holding company is buying a data collaboration and identity platform?
Is this strategic infrastructure, or just another expensive way to fight a battle that the walled gardens already won?
I am still thinking through the answer, but my current view is this:
Publicis is not buying a way to replace the walled gardens. It is buying a stronger position in the messy world outside them.
That world still matters. But it matters in a more limited, more fragmented, and more practical way than the industry sometimes likes to admit.
First, The Deal Is Not Done Yet
Small but important point: as of May 18, 2026, this is not a completed acquisition.
Publicis has entered into an agreement to acquire LiveRamp. The announced price is $38.50 per share in cash, representing about $2.167B in enterprise value and $2.546B in total equity value, including acquired net cash of $379M. The deal is expected to close before the end of 2026, subject to approvals and the usual closing conditions. LiveRamp's investor release says the same.
That distinction matters because the market, regulators, clients, competitors, and LiveRamp's existing partners still have time to react.
It also matters because LiveRamp has historically been useful partly because it could present itself as a neutral data collaboration layer. If that layer becomes owned by one agency holding company, other players will naturally ask uncomfortable questions about neutrality, pricing, governance, and data firewalls.
I am not saying those questions cannot be answered. They probably can be answered contractually, operationally, and technically.
But trust is not only a contract problem. It is also a perception problem.
And in advertising, perception has a habit of becoming strategy.
The Easy Narrative Is Too Simple
The easy narrative is:
Publicis agrees to buy LiveRamp, therefore Publicis now has a powerful identity layer, therefore Publicis can compete better against the walled gardens.
I think that is partly true, but also too clean.
The walled gardens are not strong because they have one magic ID. They are strong because they own the whole operating environment.
They know who the user is.
They know what the user did inside the platform.
They control the auction.
They control the inventory.
They control the measurement interface.
They can push advertisers toward automated buying products that make the platform itself the optimizer, the measurement provider, and the recommendation engine.
That is a very different kind of power from identity resolution across the open web.
So no, I do not think Publicis + LiveRamp suddenly becomes a true substitute for the major walled gardens. If an advertiser needs performance inside one of those ecosystems, they will still need that ecosystem. If a brand wants to appear in a large short-form video feed, marketplace search result, premium video environment, or retail media network, the platform owner still matters.
The better question is not:
Can LiveRamp help Publicis beat the walled gardens?
The better question is:
What problems do advertisers still have because the walled gardens are so strong?
That is where the deal becomes more interesting.
Why Advertisers Still Need Measurement Outside The Platforms
If you are a small advertiser spending mostly in one or two platforms, the answer may be: you do not need much else.
You can trust each platform's dashboard enough, optimize inside that platform, and move on. It may not be perfect, but it is good enough for many businesses.
But larger advertisers have a different problem.
They do not only need to know whether one platform says it performed well.
They need to know:
- Did this media actually create incremental sales?
- Did the platform claim credit for demand that would have happened anyway?
- How much overlap is there between platforms?
- What happens if we move budget from one ecosystem to another?
- What is the combined reach and frequency across premium video, retail media, social, search, commerce, and the open web?
- Are we under-investing in long-term brand effects because short-term attribution makes one channel look better?
No single walled garden can answer those questions honestly for the whole plan.
Not because people inside those companies are bad. That is not my point.
The structural incentive is the issue.
Each platform sees its own world most clearly. Each platform is rewarded when advertisers spend more inside that world. Each platform's measurement is, by definition, platform-centered.
So if an advertiser adds up the reported impact from every platform, the total can easily become more impressive than reality. Everyone gets partial credit. Sometimes everyone gets too much credit.
That is why advertisers still care about independent measurement, marketing mix modeling, incrementality testing, holdouts, clean rooms, and data collaboration.
Not because any of those are perfect.
Because the alternative is letting every seller grade its own homework.
The Open Web Still Matters, But Not In The Old Way
I think this is where the conversation gets confusing.
When people say "the open web still matters," it can sound like they are arguing that the open web will somehow return to its old role as the center of digital advertising.
I do not think that is realistic.
The old open-web identity model was built on cookies, device IDs, pixels, and a fairly casual attitude toward cross-site tracking. That world has been shrinking for years.
Safari's Intelligent Tracking Prevention changed what cross-site tracking could do. Apple's App Tracking Transparency made user consent much more explicit in iOS apps. Ad blockers, private browsing, premium subscriptions, privacy regulation, and logged-out usage all reduce what can be observed and matched. The broader browser direction has also moved toward more user choice and less invisible tracking, even though the exact path keeps evolving.
So I would be careful with any story that makes "common ID" sound like the industry is going back to 2018.
It is not.
The open web still matters, but not because it gives advertisers a universal view of everyone.
It matters because it is part of a bigger fragmented system:
- publisher environments with logged-in users
- retail media networks with purchase data
- CTV platforms with account-level identity
- commerce media partnerships
- brand CRM data
- clean rooms
- contextual signals
- aggregate measurement
That is a very different world from "drop a cookie and follow the user everywhere."
It is less complete.
It is more expensive.
It is harder to explain.
It is also probably more realistic.
AI Chatbots Make The Open Web Matter Even Less As A Destination
There is another layer here that makes the open-web question even harder.
People are increasingly getting answers directly from AI chatbots and AI-assisted search experiences. They ask a question, get a synthesized answer, maybe ask a follow-up, and often never visit the underlying websites.
I do not say this with a happy face.
I operate a website. I write long-form posts. I care about people finding the original source, reading the full argument, and building a relationship with the person behind the page. So if the open web becomes less of a destination, that affects me directly too.
But I would rather work with reality than try to negotiate with it.
The reality seems to be this: for many informational questions, the user's journey is shifting from "search -> click -> read" to "ask -> answer -> maybe click if needed."
That changes the open web's role in three ways.
First, it reduces the open web's role as a destination. The website still exists, but the user may get the answer somewhere else.
Second, it creates a new measurement blind spot. If an AI answer influences what someone believes, compares, or buys, how does an advertiser measure that influence? It is not a normal impression. It is not a normal click. It is not cleanly captured by platform attribution or open-web identity graphs.
Third, it pushes more value toward places where the data is still closer to the transaction or the logged-in relationship: retail media, CTV, commerce environments, publisher subscriptions, loyalty programs, and brand first-party data.
That matters for this deal because it makes the old "open web identity layer" story even less convincing. If users are spending more discovery time inside AI answer interfaces, then the open web is not only harder to identify across. It is also less frequently visited in the first place.
This does not make LiveRamp irrelevant. It changes what LiveRamp is relevant for.
The value is less about reconstructing a universal open-web user journey and more about helping brands, retailers, publishers, CTV platforms, and commerce partners collaborate where there is still consented, high-quality data to work with.
That is a smaller claim, and I think a more believable one.
The Common ID Problem Is Real
This is the part I wish the industry talked about more plainly.
Common IDs are useful, but they are not magic.
A match is not the same as truth.
An addressable user is not the same as a reachable user.
A reachable user is not the same as the right user.
And a user who can be measured is not necessarily representative of the users who matter most.
That last point is important.
The people who block ads, opt out of tracking, use privacy-focused browsers, pay for ad-free subscriptions, or live mostly inside closed apps are not randomly distributed. They may skew younger, wealthier, more urban, more technical, more privacy-conscious, or simply more valuable to certain brands.
So the unmeasured population is not just "missing data."
It may be systematically different data.
That creates a selection bias problem.
If your measurement mostly sees people who are easier to track, you can start optimizing toward the visible part of the market while missing the invisible part that may be more valuable.
This is why I get nervous when identity vendors talk about coverage in a way that sounds too clean. I do not doubt that many of the technical claims are true within their definitions. But the operational question is different:
How much of my actual target audience can I reach, match, measure, and trust for this specific campaign, in this specific market, on these specific channels?
That answer is usually much less elegant than the sales deck.
Some real numbers help here. LiveRamp's own documentation says match rates vary by input identifier, data quality, matching methodology, precision level, destination platform, and lookback windows, which can be as short as 30 days for some destinations. A platform may also separate "matched" from "reachable": LinkedIn's Matched Audiences docs explain that last audience count deduplicates source entries and removes members who opted out. On mobile app identity, Adjust's Q2 2025 benchmark puts ATT opt-in at 35% among users shown the prompt, which means the rate across all iOS users is lower when apps do not show the prompt. And YouGov's 48-market research found that ad-blocking and anti-tracking tools are common enough to be a real coverage issue, not an edge case.
A headline 60% match rate in one environment is not a 60% cross-web-and-app measurement truth. For some real cross-platform plans, I would not be shocked if effective addressability landed closer to 15-35% after browser limits, app consent, ad blockers, anti-tracking tools, logged-out users, deduplication, and reachability are layered in. That is not a universal benchmark. It is a planning hypothesis to test.
So Why Are Clean Rooms Still Important?
Because even a partial answer can be valuable if the alternative is no answer.
Clean rooms are not exciting because they make measurement perfect. They are exciting because they make certain types of collaboration possible under modern privacy constraints.
Retail media is the easiest example.
A retailer may know who bought the product.
A brand may know who belongs to its CRM or loyalty audience.
A media platform or publisher may know who saw the ad.
Nobody wants to leak raw personally identifiable information to everyone else. Regulators do not want that either. Clean rooms become the controlled environment where parties can match, analyze, and measure under agreed rules.
Is it complete?
No.
Is it clean in the everyday sense of the word?
Also no. Anyone who has worked with real data knows that "clean" is doing a lot of emotional labor there. :)
But in retail media, CTV, and first-party data activation, clean rooms can answer questions that would otherwise be impossible or unsafe to answer.
That is the practical value.
Not perfection.
Permission.
What Publicis Is Really Buying
So when I look at this deal, I do not see Publicis buying "the identity layer of the internet."
That sounds too grand.
I see Publicis buying several more practical things:
The scale matters here. EMARKETER forecasts US retail media spending at $69.33B in 2026, and IAB/PwC reported $63.4B in 2025 commerce media plus $78B in digital video, which includes CTV, social video, online video, and short-form video. These are not side channels anymore.
- A stronger role in retail media and commerce measurement
Retail media keeps growing because retailers have something advertisers desperately want: purchase data. LiveRamp has been important in helping brands and retailers collaborate around that data. That is strategically useful.
- A stronger role in CTV and authenticated environments
CTV is fragmented. It is not one clean garden. It is many living rooms, accounts, apps, devices, publishers, platforms, bundles, and measurement systems. Identity and data collaboration still matter there, even if they are messy.
- A deeper first-party data activation layer
Brands with real CRM, loyalty, transaction, or customer data need ways to activate and measure that data without throwing it into the open. LiveRamp helps with that.
- A defensive position as platform buying becomes more automated
If platforms keep automating more of the buying process, agencies need to justify their role somewhere else: strategy, data, measurement, experimentation, integration, and operating model. Owning more of the data collaboration layer helps.
- A story about AI agents that has an actual data substrate underneath it
Publicis is framing the deal partly around agentic transformation, and I am wary of acquisitions wrapped in the phrase of the year.
But the underlying logic actually holds. AI agents without trusted data access are mostly workflow demos. Identity, clean rooms, permissions, and data collaboration are the substrate that makes agentic media planning, activation, and measurement real rather than theatrical.
On that point, the framing makes sense to me, even if the language is fashionable.
The Risk Is Trust
The obvious risk is integration.
Buying data assets is one thing. Making them work together across teams, clients, markets, products, legal constraints, and actual campaign workflows is another thing entirely.
But the deeper risk is trust.
LiveRamp's value has depended partly on the belief that it can sit between many parties. If some of those parties now see it as owned by a competitor, they may still use it, but they will look harder at alternatives.
In effect, Publicis is paying a premium for an asset whose neutrality was part of the moat, and whose neutrality becomes harder to believe once Publicis owns it.
Publicis seems to know this concern exists. The release says LiveRamp will continue as a "neutral, interoperable platform" and that standard commercial practices will continue. That is exactly the kind of language you include when you expect clients, partners, and competitors to ask whether the asset can still be trusted after ownership changes.
They may ask:
- Are the data firewalls strong enough?
- Will pricing stay neutral?
- Will roadmap decisions favor one ecosystem?
- Can governance be audited?
- Should we diversify into other clean room, cloud, publisher, or retail media solutions?
Those are rational questions.
But this is why I do not think the deal should be understood only as "Publicis gets stronger." It may also push the rest of the market to reduce dependency on a platform that now has a different ownership structure.
That is not just a perception issue. It is a revenue-risk question for the asset itself.
In other words, the asset becomes more powerful and more politically complicated at the same time.
Both can be true.
My Practical Takeaway
If I were an advertiser looking at this from the outside, I would not change my measurement philosophy because of one acquisition announcement.
I would use it as a reminder to be more disciplined.
I would treat platform-reported ROAS as useful but not final.
I would treat common IDs as partial signals, not universal truth.
I would invest more in two buckets.
Measurement: marketing mix modeling, incrementality testing, geo experiments, holdout design, and clean room measurement where the data quality is actually strong.
Capabilities: first-party data strategy, contextual and creative quality, and retail media / CTV measurement standards.
And I would ask every identity or clean room partner a simple set of questions:
- What percentage of my target audience can you actually match?
- How much of that match is deterministic versus modeled?
- What populations are missing?
- How do you handle iOS, ad blockers, logged-out users, and premium environments?
- What decisions should I make from this data, and what decisions would be irresponsible?
That last question is the one I like most.
Good measurement should not only tell you what you can know. It should also tell you what you should not pretend to know.
Where I Land For Now
I think the Publicis-LiveRamp deal matters.
But not because it defeats the walled gardens.
It matters because the rest of advertising is becoming a set of fragmented, privacy-constrained, authenticated, partially measurable environments. Retail media, CTV, publisher data, commerce media, and brand first-party data all need connective tissue.
LiveRamp is one of the better-known pieces of that connective tissue.
Publicis is betting that owning more of it will make the group more valuable in a world where media buying itself becomes more automated and platform-controlled.
That is a defensible bet.
It is also not a magic answer.
The future of measurement is probably not one universal ID that explains everything. It is more likely a patchwork of clean rooms, aggregate models, experiments, publisher and retailer partnerships, privacy-preserving APIs, contextual signals, AI answer surfaces, and old-fashioned judgment.
Less elegant than the keynote version.
Probably closer to reality.
I might be wrong, of course. Identity and clean room specialists probably see details here that I am missing. But from where I sit, the most honest read is this:
The walled gardens still dominate. AI chatbots make the open web less central as a destination. Common ID is less reliable than the sales language suggests. And yet advertisers still need independent ways to decide where money should go.
That is why this deal matters.
Not because it gives Publicis a way around the walled gardens.
Because it gives Publicis more leverage in everything the walled gardens cannot fully answer.
That's it from me. If you work closer to identity resolution, clean rooms, retail media, or CTV measurement, I would genuinely love to hear where you agree or disagree.
Still figuring it out.
Cheers, Chandler