The IAB published its first serious framework for measuring how brands and publishers show up in AI search — and spent almost as much effort explaining what the document is not. “Measuring Visibility in the AI Era” proposes a four-level hierarchy, the 4Ps: Presence (how often you are cited at all), Prominence (where you appear, and whether you are highlighted or buried in a list), Portrayal (sentiment and factual accuracy — including whether the AI is hallucinating things about you), and Persuasion (whether a citation ever becomes a click-through). It is, the IAB stresses, explicitly not a standard.
That refusal is deliberate, and it is the most informative thing in the document. IAB VP of AI Caroline Gigerenich says AI search simply lacks the consistency to standardise yet — the platforms’ behaviour shifts too often and varies too much for any fixed measurement spec to hold.
What the guidance does offer is a pair of practical lines that publishers should tattoo onto their measurement practice. First: a minimum of 50 queries before visibility data characterises anything — below that threshold, the IAB’s own language is that the data is exploratory. Second: a hard distinction between “directional” and “decision-grade” measurement, and an instruction to know, at all times, which one you are looking at.
01Procurement ammunition, delivered on schedule
The timing matters because of what is currently landing in every publisher’s inbox. The AI-visibility vendor market is in its land-grab phase, selling hard into publisher anxiety about AI search — dashboards that score your “AI presence,” often built on a couple of dozen prompts run against a couple of platforms, refreshed on an interval the pitch deck does not specify. Until this week, pushing back on those pitches meant arguing methodology from scratch. Now there is an industry-body citation to attach.
Every vendor pitch now gets three questions with the IAB’s name on them. How many queries underpin this score — because anything under 50 is exploratory by the IAB’s own line, and a score built on a dozen prompts is an anecdote with a dashboard. Is this directional or decision-grade — and will you put that word in the contract? And which of the 4Ps does it actually measure — because most tools on the market today count Presence only, while the commercial stakes sit higher up the hierarchy. Portrayal is where reputational risk lives: an AI system confidently mis-describing your brand, your content or your commerce recommendations is a materially worse problem than not being cited at all. Persuasion is where the revenue question lives: whether a citation ever becomes a visit. A vendor measuring neither is measuring the cheap layer and charging for the expensive one.
The framework earns its keep internally too. “AI visibility” is currently a phrase that means something different in every meeting — audience teams report it, executives ask about it, nobody defines it. The 4Ps supply a shared vocabulary even without being a standard: when your team reports AI visibility, make them say which P, on how many queries, refreshed how often. The discipline sounds pedantic; it is actually the thing that keeps exploratory numbers out of board decks, where they otherwise calcify into targets and then into strategy.
02Read the refusal itself
There is a second-order message in the IAB declining to standardise, and publishers should take it at face value: the platforms’ behaviour is still too inconsistent to measure reliably. Follow that logic one step further — behaviour too inconsistent to measure is far too inconsistent to build strategy on. A citation pattern observed in August can be rearranged by a model update in September, silently, with no changelog. That argues for a specific posture: continuous, cheap, methodologically honest monitoring — and a refusal to make expensive commitments (content restructuring, markdown pipelines, vendor lock-ins, headcount) on the strength of numbers the industry’s own trade body classifies as exploratory.
The most useful thing the IAB shipped this week is not the framework — it is the epistemology.
03Why this matters for publishers
| You now have a citable floor for measurement quality | The 50-query minimum and the directional/decision-grade distinction convert vendor-pitch scepticism from an opinion into an IAB-referenced procurement criterion. |
|---|---|
| Most tools measure the cheapest P | Presence counting is commoditising fast; Portrayal (is the AI describing you accurately?) and Persuasion (does a citation become a visit?) are where risk and revenue actually sit, and where tooling is thinnest. |
| Exploratory numbers calcify if you let them | A directional score that reaches a board deck unlabelled becomes a KPI by next quarter. The vocabulary discipline is the defence. |
| No standard means no stable ground | The IAB is telling you the terrain still moves. Measurement should be continuous and cheap; strategic commitments keyed to AI visibility should be small and reversible. |
04What publishers should do
05The bottom line
The most useful thing the IAB shipped this week is not the framework — it is the epistemology. AI visibility is measurable enough to monitor and too unstable to standardise, which means it is too unstable to bet strategy on. Publishers should take the free procurement leverage, install the vocabulary, run a cheap owned baseline — and hold their nerve. In a market where vendors are selling certainty about a surface the industry’s own trade body will not yet standardise, the operators who measure monthly and commit to nothing will be the ones still standing when the terrain finally settles.