Ad Tech Briefing: AI advertising’s measurement problem is really a governance problem

The advertising industry has spent years rebuilding measurement for a world with fewer identifiers. Now it has another complication: artificial intelligence is changing not only how ads are delivered, but potentially who — or what — encounters them before a purchase takes place.

For precisely this reason, the ⁠Interactive Advertising Bureau is developing a framework for measuring advertising served to, or influenced by, AI agents, with initial results expected in November.

The initiative aims to establish common ground on questions conventional attribution was never designed to answer. For example, if an AI agent researches products, encounters commercial messages and ultimately helps its user make a purchase, then the industry needs to determine what constitutes an advertising exposure, which signals demonstrate influence and how credit should be allocated across the ecosystem.

While these questions may seem highly theoretical, AI platforms were largely viewed as emerging consumer interfaces; they are becoming more commercially significant as those same companies build advertising businesses, and there is no doubt the biggest names in the industry will aim to move the needle in a direction that benefits them.

The reason why? Well, that’s simple: we need only look at the speed of ⁠OpenAI’s expansion to illustrate the point. Six months after introducing advertising in the U.S., OpenAI has expanded ChatGPT Ads across multiple international markets and is now bringing the business into another 31 European markets. Along the way, it has developed buying tools, introduced cost-per-click pricing, recruited experienced advertising executives and built relationships with agencies and ad tech partners.

That makes the IAB’s work increasingly relevant. Advertisers accustomed to measuring search, social and display media will want comparable evidence that spending inside AI environments contributes to business outcomes. Publishers will similarly want mechanisms capable of recognizing when their content informs an AI-generated recommendation or commercial decision.

Creating standards, however, is only part of the problem. The more consequential question may be who controls the underlying signals.

If AI platforms retain the data required to demonstrate how their systems influenced a transaction, independent measurement could ultimately depend on information supplied by the companies selling the advertising. Industry frameworks could set definitions and expectations, but their effectiveness will depend on whether platforms provide sufficient access for those standards to be applied independently.

If some of you are thinking: haven’t I heard this before? Then you’re right…

A recent example is the ongoing ⁠debate surrounding the World Wide Web Consortium’s proposed Attribution API, which offers a parallel example. Its supporters see browser-mediated attribution as a necessary attempt to preserve advertising measurement while limiting cross-site tracking. Critics have questioned whether shifting more attribution responsibility to browsers risks concentrating influence among the companies controlling those environments.

The circumstances are different, but the governance question is not, with debates over how standards emerge through organizations such as the IAB, measurement, or technical standards bodies often deemed proxy wars by many.

And participation alone does not guarantee a level playing field, as large technology companies generally have more technical resources, data, and personnel to participate in lengthy standards processes than smaller publishers, independent ad tech companies, or advertisers.

Many will agree the need for industry-led standardization is unquestionable. However, in both cases, common specifications could reduce fragmentation and give advertisers and publishers clearer rules for operating in rapidly changing environments. Indeed, OpenAI’s rapid commercialization suggests the AI advertising market may need those rules sooner than expected.

But establishing measurement standards will be only the first test. The next will be whether the resulting systems can provide meaningful transparency and interoperability without simply transferring measurement power from one generation of dominant technology platforms to another.

What we’ve heard

“Most on the buy side want to get closer to the impression; DSP/SSP are already converging at a rapid rate, with both building features to replace each other, all while ad servers are building direct on-ramps for execution. Where will agencies point their buying agents?! I would prob follow that puck.”

– An unnamed Digiday source gave their take on where the smart investors’ money is going.

Numbers to know 

Google’s AI workload acceleration:

• 9.7 trillion: AI tokens processed across Google’s platforms in May 2024.
• 3.2 quadrillion+: AI tokens processed across Google’s platforms in May 2026.
• Nearly 7X: Year-over-year growth in AI tokens processed as of May 2026, reflecting rapidly increasing demand for AI inference.

What we’ve covered

Stagwell’s Curate launch puts agencies’ ad tech ambitions in focus 

Earlier this month, the holdco launched Stagwell Curate, its own inventory marketplace… but that’s not to say Stagwell’s offering is meant to replace ad tech — at least if you ask Matt Adams, global CEO of Stagwell Media Platform. 

The Trade Desk is betting part of its recovery on measurement nobody can argue with

The Trade Desk missed its second-quarter forecast. Advertisers buying on price, not value, are a big part of the reason why. So now it’s betting on measurement so clear they can’t keep doing it.

What we’re reading

Where media money wants to go

Think of advertising not as a tool of marketing but as a means of capital allocation. From here, Tom Triscari argues, it’s easier to understand why capital flows toward platforms like Google, Meta, and Amazon, which provide both superior outcomes and low-friction deployment, rather than to transparent but fragmented open-web ecosystems.

A new Bill targets ‘bad bots’ that scrape websites without permission

In late July, the Stealth Bot Prohibition Act was introduced in the House, with the bipartisan legislation authored by the News/Media Alliance, in a bid to require AI stealth crawlers to disclose their identity and purpose to the host of a website – similar to legislation passed in New York earlier in the year. 

Advertising salary guide

Ad Age shows readers how to land an evolving, high-paying, and in-demand business development job.

LiveRamp shareholder approve $2.2 billion deal, but reject execs’ $82.6 million payday

Shareholders’ disapproval of proposed “golden parachute” provision won’t bar LiveRamp execs from a big payout