In the space of three weeks, the IAB has published a visibility-measurement playbook, promised an attribution framework for November, and watched three different trade outlets describe the same rollout in three different ways. Read together, the coverage says something none of the individual pieces do: the industry is not standardizing AI measurement so much as improvising it, one guideline at a time, while the people writing the rules openly admit they are not rules yet.
What actually happened
On August 3, the IAB released “Measuring Visibility in the AI Era,” a playbook built around what it calls the 4 P’s: Presence, Prominence, Portrayal and Persuasion. It answers a narrow but real problem. More than 20 vendors now sell AI-visibility measurement tools, and Marketing Dive reported that those tools routinely disagree on the same brand’s numbers. The playbook also splits results into two tiers: “directional” data for spotting trends, and “decision-grade” data, which is supposed to be the only kind rigorous enough to justify a budget call.
Three weeks later, the IAB confirmed it is building a second, harder piece: a framework for attributing paid ads served to AI agents, due November 12. Digiday reported that the framework will try to separate an AI system’s influence into two layers, awareness or intent versus the actual decision, and will have to settle who gets credited when an AI agent reads a product page, compares it to competitors, and buys on a user’s behalf without the person ever seeing a display ad.
Where the coverage lines up, and where it splits
All three outlets agree on the underlying fact: the industry’s existing measurement stack, built around clicks, UTM parameters and referral logs, does not survive an AI intermediary. Where the accounts diverge is in what the guidance means.
Marketing Dive framed the playbook as urgent infrastructure, citing McKinsey research that laggard brands could see traffic declines of up to 50% from traditional search, and noting that only 16% of brands currently track AI visibility in any systematic way. In that framing, the IAB is a standards body finally catching up to a channel that already matters.
AdExchanger told a more cautious version of the same story. Caroline Giegerich, the IAB’s VP of AI, told the outlet the guidance is deliberately not a standard: a standard requires stability, and the market is still in what she called a “mass transition space.” That is a meaningfully different claim than “the IAB sets a measurement standard,” which is how some coverage of the same release characterized it. The gap between those two framings, official guidance versus enforceable standard, is the story most accounts undersell.
Digiday’s reporting on the November attribution framework shows why that caution is warranted. Interviewed sources are not close to agreement. Jaime Schultheis of Bombora told Digiday that big AI platforms “have grown audiences off publishers with very little reciprocation,” a direct claim that publishers are being disintermediated without compensation. Michael Bishop of OpenAds countered that AI platforms function as “black boxes,” meaning any credible attribution will depend on integrations the platforms themselves control. Those are not two views of the same solution; they are two different ideas about who holds the leverage, and the framework due in November has to resolve both before it can ship.
The gap the IAB is trying to close
What makes the November framework hard is that vendors are not waiting for it. In a separate briefing, Digiday documented three companies already testing three incompatible ways to prove an ad influenced an AI agent. OpenAds embeds unique referral codes in AI-generated responses and checks whether a code, say a 10% discount, survives an agent’s retrieval and gets used. Oasy injects text-based sponsored placements directly into the HTML that AI crawlers scrape from publisher pages, then tries to track cost-per-click in a channel where no human ever technically clicked anything. Time built FAQ-style sponsored placements into its markdown pages and measures success as a shift in AI-generated brand favorability scores rather than a conversion at all.
None of those three methodologies would produce comparable numbers if placed side by side, which is precisely the fragmentation problem the IAB says its own visibility playbook exists to fix. Steven Liss, OpenAds’ co-founder, told Digiday that advertisers will not move past “experimental budgets” until someone can show them “some kind of outcome,” and Oasy co-founder Choy Travers put the underlying problem more bluntly: in AI visibility, “there is no exact science, unlike Google Ads.” That is a candid admission from two of the vendors building the measurement layer the IAB is trying to standardize, arriving in the same week the IAB confirmed its own attribution framework is still three months from release.
What the pattern actually adds up to
Taken individually, each piece of coverage reads as routine trade-press process journalism: a group publishes a playbook, a framework gets a release date, a few executives are quoted. Read as a set, the pattern is that the IAB is running its AI-measurement rollout in public, in phases, and each phase is arriving before the last one has proven itself. The visibility playbook is barely three weeks old and has no independent auditor checking whether vendors actually apply its 4 P’s consistently, a gap this publication has already flagged in the context of Nielsen’s own self-graded bias corrections. The attribution framework due in November is being built while its two most vocal stakeholder camps, publishers and platform operators, are still arguing about who the framework should protect.
That sequencing matters for a martech leader more than any single guideline does. A framework built to referee a fight that has not been settled yet is not a finished standard a team can build reporting around. It is a draft position paper that happens to have a launch date.
What it means for the marketing leader
Treat both IAB releases as directional, not decision-grade, for now, which is in fact the IAB’s own language. Do not retire existing attribution models in favor of a framework that has not shipped, and do not let a vendor claim compliance with “the IAB standard” when the IAB itself says no standard exists yet. For visibility tracking specifically, ask any vendor whether its numbers meet the IAB’s own decision-grade bar, query volume, sample size and reproducibility, before using those numbers in a budget conversation. For paid AI-agent placements, the more useful question right now is contractual, not statistical: what does the platform commit to disclosing about how an ad influenced an agent’s decision, since that disclosure, not a November framework, is what any attribution model will eventually have to be built on.
It is also worth watching who is not yet in the room. Publisher representatives like Bombora are pressing the reciprocation argument loudly because the November framework’s early drafting has, by Digiday’s account, been dominated by platform and vendor voices. A marketing leader whose media plan depends on publisher inventory has a direct stake in how that argument resolves, since a framework that credits only the AI platform’s own signals will systematically undercount the publisher content that made the AI response possible in the first place. That is a commercial question worth raising with any AI-ads vendor now, months before the framework locks in an answer nobody outside the drafting room gets to vote on.
The honest read of this month’s coverage is that the rulebook for measuring AI’s role in marketing is being written in real time, by a standards body that keeps saying, correctly, that it is not done yet. Budgeting and reporting decisions should reflect that, even when a vendor’s pitch deck does not.
Source: IAB
