The trade body that brought order to digital advertising measurement has turned to AI visibility, and its opening move is to tell marketing teams which of their numbers are not good enough to spend against. The IAB published Measuring Visibility in the AI Era on 3 August, sorting the data into two grades: directional, which is useful for spotting patterns, and decision-grade, the only tier the framework considers fit for budget allocation or executive strategy.

The category outgrew its methods before anyone agreed what to measure. More than 20 companies now sell AI visibility measurement, each using a different approach, and the same brand can receive different answers depending on which it buys. Marketing teams are already spending against those answers. About 24% of search or content budgets now goes to AI visibility, according to a survey by content agency Fractl, and 73% of marketers have bought monitoring tools, per a survey from Scrunch and Scribewise, though Scrunch sells one of those tools. What the framework supplies is a way to ask whether the figure in the deck can carry the decision resting on it.

What does the IAB framework standardise?

The IAB has set a shared vocabulary, quality criteria and disclosure requirements without ranking providers or prescribing tools. Its metrics hierarchy runs in four stages the bureau calls the four Ps. Presence asks whether a brand appears at all, through mention rate, citation rate, share of voice and visibility momentum. Prominence asks where it appears, covering placement, ranking order and whether content is drawn on substantively or cited in passing. Portrayal asks in what context and with what accuracy, through sentiment, framing, hallucination rate and factual inaccuracy rate. Persuasion asks whether any of it drives action, through recommendation strength and post-citation click-through rate.

Portrayal is the stage with no equivalent in older measurement. Hallucination rate and factual inaccuracy rate treat a wrong answer about a brand as a measurable brand-safety problem rather than an anecdote, which matters in a year when a Munich court ruled that Google’s AI Overviews are the company’s own statements and can be actioned as such.

The quality classification is the part a marketing manager can use this week. Directional data identifies patterns and supports early signal detection and competitive awareness, but the framework states plainly that it is not sufficient for budget allocation or executive strategy. Decision-grade data has to clear a higher bar on query volume, sample size, prompt type coverage, testing cadence, reproducibility and platform coverage before it is used to move money. Caroline Giegerich, the IAB’s vice-president for AI, said measurement frameworks have not kept pace with how people now discover brands inside AI platforms.

Why can’t the tools show revenue?

Scrunch, Profound, Semrush and the rest of the category measure presence in an answer, not what happens after it. They can estimate how an assistant describes a brand and which sources it draws on. They cannot follow the buyer who reads that answer, opens a tab a week later and arrives through a branded search, which is where a large share of AI-influenced traffic lands.

Rippling shows what closing the gap currently costs. The B2B software company works back from visibility measures taken from Profound and AirOps, conversion data from paid ChatGPT ads, traffic from branded and unbranded search queries, and a bespoke media mix model built on Meridian, Google’s open-source MMM platform, its head of growth Neel Murthy told Digiday. Performance media agency Roast has been running Google’s CausalImpact model, which uses Bayesian inference to connect search inputs to business outcomes. John Barham, a managing partner at Roast, argues that search now has to be understood in probabilistic terms rather than as a channel with a downloadable report at the end of it. Both approaches need in-house data science, which is the part no marketing team can buy off a shelf.

Adoption of the cheaper measures is thinner than the spending suggests. In the Scrunch and Scribewise survey, 71% of respondents were not tracking share of voice against competitors, 70% were not monitoring sentiment towards their brand and 67% were not analysing AI bot traffic. Those are Presence and Portrayal metrics in the IAB’s own hierarchy, which puts most teams several stages short of the Persuasion questions they are being asked to answer. “Citation alone isn’t good enough,” said Andrew Wheeler, chief executive of content marketing agency Skyword, who argued that authority has to be present at the moment of citation and then convert into a business outcome.

How does the standard read the numbers already in circulation?

The 303% rise in ChatGPT referrals to B2B sites that Demandbase reported in August is a useful test of the classification. Run against the IAB’s criteria, it is strong where the framework asks for sample size, drawing on more than 11 billion website visits across 1,584 Demandbase instances. It is weaker on platform coverage and on the base being measured, which is one vendor’s customer set rather than the B2B web, and weaker again on reproducibility.

That last point has sharpened since publication. Demandbase placed the steepest jump in May 2026 and offered no explanation for it. Similarweb has since reported that after 7 May 2026, when ChatGPT began surfacing clickable brand links inside answers instead of in footnote-style citations, total ChatGPT referrals rose 157.7% week on week and referrals landing on brand homepages rose 354.7%, with the homepage share of ChatGPT clicks moving from roughly a quarter to around 60% and holding there. Neither company links the two observations, and neither dataset establishes that they describe the same change.

A figure that can double on an undisclosed interface decision is directional by the IAB’s own definition, whatever its sample size. That is not a reason to ignore it. It is a reason to present it as a trend rather than as the basis of a forecast, and the framework now gives that distinction a name a finance director will accept.

The IAB places the link between visibility and action in its Persuasion tier, which it says bridges to a forthcoming attribution framework. That framework has not been published, and the IAB has not given a date for it.