IAB Admits AI Attribution Cannot Be Standardized: Publishers Lack Data

The Interactive Advertising Bureau published its "AI-Era Attribution Blueprint" today after four months of work, and the document concedes the industry's core problem: conversions influenced by AI cannot be reliably measured because the data sits inside AI platforms that advertisers and measurement providers cannot access.

Digiday first reported in August that the IAB was developing an AI attribution framework. Its working group included Walmart, WPP Media, Criteo, Microsoft, Tinuiti, Prudential Financial, People Inc., Emberos, Bluefish and Brandlight. The blueprint classifies who or what created intent, made the decision and executed the transaction; defines the claims AI involvement creates and the evidence each requires; and highlights the transparency gaps that block verification.

"Going in, I thought that we were going to help figure out how credit was assigned through the process. I realized actually what this document was starting to become was more of a call to action to the industry," said Caroline Giegerich, the IAB's vp of AI, during a virtual roundtable on Monday. "There are evidence layers that are missing."

What do the numbers show?

The attribution gap is measurable, and it skews budgets toward channels that capture conversions rather than those that influence them. Similarweb data cited in the IAB report found consumers who receive a brand recommendation from an AI assistant are 2.5 times more likely to visit that brand's website within seven days. Only 8.8% of those visits register as AI referrals. Instead, 55.9% are attributed to search and 19.9% to direct traffic.

The clearest distortion hits publishers. If a consumer asks an AI answer engine for the best product and the answer draws on a publisher's review, the consumer may never click the citation, and the retailer logs it as a search or direct sale. "The review shaped the sale, and the publisher is eligible for a contributory claim," the report reads, but content used in model training is generally not traceable from outside the model.

"Publishers are feeding the answers but they don't have the receipts," said Jaime Schultheis, head of global data partnerships at Bombora.

The IAB recommends platforms disclose whether recommendations are paid or organic, name the sources informing them and label AI-originated traffic consistently. Brands and agencies should apply the framework's claim-and-evidence standards across platforms; publishers should identify claims about their content's influence and gather supporting evidence.

What does this change for practitioners?

For now, publishers are patching the gap vendor by vendor. Time works with Mobian to measure ads served to AI agents. Bustle Digital Group uses Profound to measure GEO campaign performance for brands. These efforts show specific activity but do not tie AI influence to sales.

Jon Roberts, chief innovation officer at People Inc., said the framework's value is comparative. "Everyone's pitching you on their custom proprietary measurement solution, and they're all using the same terms, and the terms all mean something different," he said during the roundtable.

Relief may come from two directions by a fixed date. The U.K.'s Competition and Markets Authority requires Google to provide U.K. publishers with clickthrough data for its AI search features by Dec. 3. OpenAI is testing an analytics dashboard that counts how often a business appears in ChatGPT's answers and how many clicks those appearances send. "We're waiting [with] bated breath," said Graham Howes, svp of distribution and AI intelligence at BDG.

Brands face the same constraint. Ross Bucholc, associate media director at Prudential, said the lack of first-party data from AI platforms is holding back additional ad budget for LLMs.

Even with visibility data, connecting AI activity to purchases remains unsolved. Profound's analysis of browser-session data found one in three consumers had a relevant AI conversation about their purchase within a month that went unattributed, though the dataset excluded Google's AI Mode and AI Overviews. "It's a really tricky problem right now," said Josh Blyska, Profound's head of insights.

The Guardian's head of sales for North America, John Sills, framed the demand plainly: "These LLMs are in the ad business now... we need that level of account management to help guide the best practices."

Elsewhere this week

• Meta indefinitely paused terms of service changes that would have allowed scraping publishers' websites, after backlash and regulatory scrutiny, Nieman Lab reported.
• Yahoo is expanding Scout, its AI-powered search experience, across News, Finance and Sports, per Fast Company.
• The Online News Association canceled its 2027 conference and paused all in-person events, citing rising costs and shrinking sponsorships.
• Press Forward has invested nearly $600 million in news organizations, two years ahead of schedule, the AP reported.
• Anthropic struck its first news licensing deal, with the Associated Press, to provide real-time election data in Claude, Adweek reported.
• The New York Times Company is considering acquiring film community Letterboxd for a reported $330 million.
• Ziff Davis content was cited 2.7 billion times in Google's AI Overviews and AI Mode in Q3 — nearly 10 times its Google Search clicks — even as referrals fell 37% year over year.
• Washington Post executive editor Matt Murray said 95% of the outlet's AI referrals now come through OpenAI, telling Web Summit: "That's a win for them and a win for us."

Until platform click data arrives and third parties can verify it, the blueprint functions less as a standard than as a documented demand for evidence the industry does not yet possess.