The ad market is forecast to cross the trillion-dollar milestone for the first time in 2026. What it cannot do is measure where that money actually went.
For the brands and agencies I speak with, that is the real problem of the moment. Budgets are consolidating, but the data needed to deploy them intelligently keeps scattering across retail media networks, connected TV platforms, and walled gardens that were built to retain signal, not share it. Every new channel adds another pool of data that does not talk to the rest.
As AI raises the stakes on both sides of that wall, I’ve come to see secure data collaboration as infrastructure rather than as a marketing tactic. It is the layer that decides whether any of this connects.
At a glance
● Budgets are concentrating; the data to deploy them is scattering.
● Signal loss and fragmented data environments are now tied directly to a measurement-confidence problem the industry openly acknowledges.
● Privacy-preserving data collaboration lets two or more parties analyze combined data without either one exposing its raw records.
● Treated as infrastructure, it becomes a connective layer you build once and reuse across every partner, channel, and measurement deal: bringing intelligence to the data rather than moving data to the intelligence.
Why advertising and measurement keep fragmenting
Contrary to what many vendors would have you believe, fragmentation isn’t simply the result of a lack of tools. It is the structural direction of the market, and two forces are driving it:
Signal loss and the pull of walled gardens
Third-party identifiers are disappearing, and the identity solutions replacing them do not interoperate. Each platform offers its own login, its own graph, and its own measurement. The richest audience and outcome data now sits inside environments designed not to share. Meanwhile, the open web faces the opposite problem: too many fragmented identity solutions competing to replace what cookies once unified, with no clear interoperability between them.
Marketers feel this most at the measurement layer. The IAB’s 2026 State of Data report puts it plainly: “privacy regulation, signal loss, platform-embedded optimization, and fragmented data environments have made it harder to connect media exposure to outcomes with confidence.”
AI is compounding the problem. As models get embedded deeper into platform infrastructure, the incentive to retain proprietary data grows stronger. The signal that used to leak out through third-party cookies is now being absorbed into closed systems that learn from it. The walled gardens continue to hold data while also now actively training on it.
That dynamic plays out at every level of the supply chain. The brand holds its first-party data, the publisher holds the exposure data, and the retailer holds the purchase data. None can be combined safely if one party has to hand raw records to another, which privacy law and commercial caution both forbid.
Retail media and CTV multiply the silos
New advertising channels are increasingly “walled” from the outset. Retail media networks and connected TV are among the fastest-growing parts of the market, and each one is its own measurement island.
The concentration is striking: eMarketer projects that by 2028, Amazon’s retail media revenues will exceed $75 billion, more than $65 billion ahead of the next-largest network. Walmart’s acquisition of Vibe.co signals the same dynamic extending into connected TV, as the largest retail media players move to own measurement across channels as well as inventory. Each network reports outcomes in its own closed environment, on its own terms.
For an agency running a national campaign, that means reconciling results across a dozen platforms that were never built to agree. The brand sees spend in one system and outcomes in another. Proving what actually worked turns into guesswork dressed up as attribution.
The case for treating collaboration as infrastructure
Here is the shift I would urge brands and agencies to make. If the need to combine data with another party is occasional, you can treat it as a project: a one-off integration, a bespoke contract, a single data pull. If the need is constant and spread across many partners, it stops being a project and becomes an infrastructure question.
That is where the market sits today. A brand does not collaborate with one retailer once; it works continuously with several retail media networks, multiple publishers, a measurement partner, and its own agency (for example).
Building a fresh, ad-hoc data arrangement for each of those relationships is slow, legally fraught, and impossible to scale.
Infrastructure is what you build once and reuse, independently of any single platform or provider. Secure data collaboration, done as infrastructure, is then a standing capability: a neutral layer where any approved partner can run joint analysis under rules agreed in advance, with raw data never changing hands.
What changes when collaboration is the layer instead of the project
When secure data collaboration becomes infrastructure rather than a string of one-offs, a few key things change for brands and agencies. New partnerships move and scale faster, because onboarding becomes a permissioning decision rather than a three-month legal and engineering build.
Privacy becomes more trustworthy, because it is enforced by how the system is built rather than promised in a contract (which is especially crucial once agents come into play). And measurement improves, because exposure, spend, and outcome data can be joined across partners to produce attribution that holds up, without anyone surrendering the underlying data they depend on.
This is the model behind Decentriq: a neutral layer where any approved partner can run joint analysis under rules agreed in advance. It is why we treat secure data collaboration as advertising infrastructure rather than a feature.
How to tell infrastructure from a workaround
When I assess an approach, a few signals separate the two:
● Neutrality: The layer sits between the parties and favors none of them, the operating vendor included.
● Enforcement over promises: Privacy rests on how data is processed, for example hardware-level isolation that keeps raw records invisible to every party, rather than on assurances written into a contract.
● Reusability: One setup serves many partners and use cases.
● Auditability: Every party can verify what was run and what was allowed to leave the environment.
● AI and agent readiness: The layer is built to support automated, model-driven workflows, including agentic systems acting on behalf of buyers or sellers, without compromising the privacy guarantees that govern every other interaction within it.
A workaround solves this quarter’s campaign. Infrastructure solves the next hundred. If an approach cannot be reused across partners, or cannot show you what happened inside it, it is the former.
Questions brands and agencies are asking about secure data collaboration
Why does secure data collaboration matter now?
Secure data collaboration matters now because the data needed to plan and measure advertising has scattered across platforms, retailers, and devices that will not share raw records. It is the practical way to combine that data lawfully and usefully, so brands and agencies can target and measure across partners without exposing what any single party owns.
How do we improve collaboration across partners without exposing data?
Improvement starts with making collaboration infrastructure: one neutral, privacy-enforcing layer that every approved partner uses, rather than a bespoke arrangement built fresh for each relationship. Standardize that layer once, set the rules in advance, and each new partnership will no longer require months-long discussions and agreements before starting.
Build the layer before you need it
The brands and agencies that spend well over the next few years will not be the ones waiting for fragmentation to reverse. It will not. They will be the ones that treat secure data collaboration as something they build deliberately, ahead of the campaign that depends on it.
My advice is straightforward: stop scoping data collaboration deal by deal. Instead, decide what your standing layer looks like, which partners it has to connect, and what privacy protections it must technically enforce, then build it once.
Fragmentation is not going to reverse. That is exactly why the response to it should be infrastructure rather than a workaround. The brands and agencies that figure that out now will not be scrambling to catch up when the next channel fragments things further.
Maximilian Groth:
Maximilian Groth is the co-founder and CEO of Decentriq, a technology company founded in Switzerland. The company specializes in secure data collaboration and offers a platform for data clean rooms, as well as the Collaborative Audience Platform: a unified layer that adds CDP- and DMP-style capabilities to the clean room for real-time segmentation, identity, activation, and shared audience products. Decentriq has secured significant funding, acquired international customers, and established partnerships with major technology companies such as Microsoft.
AdTech Edge is a leading digital publication covering the latest developments in advertising technology, AI-driven marketing, programmatic advertising, retail media, connected TV (CTV), data privacy, media buying, and digital advertising innovation.
