If you are a CMO, marketing director, or analytics lead who has been handed an AI visibility dashboard and asked whether the numbers on it can move budget, the IAB just gave you a ruler. Almost nobody is using it. This post is the twenty-minute test you can run on your own vendor, today, without a statistician in the room.

TL;DR

• The IAB drew a line on August 3. Its guidance, Measuring Visibility in the AI Era, sorts AI visibility data into three tiers: exploratory, directional, and decision grade. Only the third tier is fit to allocate budget against.

• Under 50 queries is exploratory. The IAB says that volume cannot meaningfully characterize a category. If your vendor tracks 40 prompts, you own a conversation starter, not a measurement program.

• Directional is not a soft version of decision grade. It is a different product with a different job: early signal, internal briefings, competitive awareness. It is explicitly not for budget allocation.

• Twenty-plus vendors, twenty-plus methods. The IAB counted more than 20 companies selling AI visibility measurement, each on its own approach. Only 16% of brands track AI visibility at all.

• Score your own tool in twenty minutes. The 12 questions below map directly onto the tier definitions. Six answers put you in a tier. The other six tell you how far you are from the next one.

We participated in the IAB working group discussions, and we have written elsewhere about what the 4 P’s mean for a 2026 measurement stack. This post is not that. We are not going to argue for the standard here. We are going to hand you the ruler and get out of the way.

What the IAB actually published on August 3

Measuring Visibility in the AI Era is guidance, not a ratified standard, and the IAB is careful about the difference. Caroline Giegerich, the IAB’s VP of AI, framed the gap plainly: consumers are increasingly discovering and considering brands inside AI platforms, but measurement frameworks have not kept pace. She has also said the thing that would keep her up at night is factual inaccuracy, which is the part of the guidance most marketers skim past.

The document does two useful things. It proposes a shared metrics hierarchy, the 4 P’s: Presence, Prominence, Portrayal, and Persuasion. And it sorts the data underneath those metrics into tiers by how much decision weight the data can carry.

The working group was not a vendor huddle. Walmart, Acxiom, Microsoft, WPP Media, eMarketer, and Tinuiti were all in it. That matters when you take this into a vendor conversation, because the tiers are buy-side language, not seller language.

The three tiers, in the language your vendor will not use

Here is the whole framework, translated.

Exploratory. Fewer than 50 queries in the measurement program. The IAB’s position is that this volume cannot meaningfully characterize a category. Useful for a hunch. Not useful for anything with a dollar attached.

Directional. Enough to answer trajectory questions. Fit for early signal detection, internal briefings, and competitive awareness. The guidance is explicit that directional data is not sufficient for budget allocation. That word, “sufficient,” is doing a lot of work, and it is the word your vendor will not repeat back to you.

Decision grade. Clears a higher bar across eight things at once: sample size, query volume, prompt-type coverage, testing cadence, reproducibility, and documented method. Fit for budget allocation, provider selection, and executive strategy.

Notice what is missing from all three definitions. Engine coverage. Pricing tier. Integration count. Dashboard design. Those are table stakes in 2026, common to every vendor in the category, and the IAB did not use a single one of them to define a tier. Neither should you.

The coin flip that explains why the tiers exist

Think of flipping a fair coin 50 times. Sometimes you’ll get 23 heads. Sometimes 27. Sometimes 30. Same coin. Just different runs. The variation isn’t because the coin changed, it’s because 50 flips is a small sample, and small samples bounce around a lot. AI visibility measurement at 50 prompts works the same way.

That is the entire argument for tiering, and it is why the IAB drew its floor at 50. We have watched the same brand sit inside an answer at 9:00 AM and vanish from the same prompt by 9:05. Sometimes the difference between two runs is cosmetic. Often it is not. The engine did not change its mind about your brand. You just took two draws from a distribution that was always wider than the dashboard admitted.

Once you see the tiers as a statement about how wide that distribution is, the twelve questions below stop feeling like a compliance checklist and start feeling like the only questions worth asking.

The 12-question scorecard

Score each one Yes or No. No partial credit, and no credit for a roadmap promise. Demo floor only: if the vendor cannot show it to you in the product, it does not count.

Questions 1 and 2 are the gate. Everything from 3 down assumes you cleared it.

How to score it

Nine or more Yes, including 3, 5, 8, and 9. Decision grade. Move budget on it. Ask for the method document in writing anyway.

Five to eight Yes. Directional. Genuinely useful for a Monday standup and a competitive read. Take it out of the board deck, and take it out of the reallocation conversation.

Four or fewer, or a No on question 1 or 2. Exploratory. Whatever else the product does well, the number on the screen is not carrying the weight you are putting on it.

Most tools we have audited land in the five-to-eight band. That is not an insult. Directional measurement is a real product with a real job, and a lot of the category builds it well. The failure is the labeling, not the tier: directional data sold with decision-grade confidence, wrapped in a case study, and put in front of a CMO.

I was slow to see this myself. For most of 2025, I thought the category’s problem was engine coverage, and my co-founder Todd and I said so more than once. Coverage was never the constraint. More prompts alone will not fix it either, and neither will faster dashboards, broader engine lists, or overlaying agentic workflows or recommendations on top of a number that was never stable or accurate to begin with.

Marketer-Side Impact

Take a mid-size content program at roughly $750 per briefed piece. A single quarter’s reallocation off a directional reading, say 60 pieces pointed at the wrong query cluster, is about $45,000 of production spend chasing a number that could not tell movement from noise. That is before agency rework, before the credibility cost of walking a number back in a QBR, and before the client d/sat that follows when the second run does not reproduce the first. Across the category, we estimate the industry burns roughly $2 billion a year acting on AI visibility numbers that do not hold up. The IAB’s own count says only 16% of brands are tracking this at all, which means most of that waste is still ahead of us, not behind us.

Three questions for your next vendor call

• “Which IAB tier does this dashboard sit in, and which specific criteria put it there?” A vendor who has read the guidance will answer in about fifteen seconds. A vendor who has not will pivot to engine coverage.

• “If you re-ran my exact program tomorrow with no changes on my side, how much would this number move?” There is a correct answer, and it is a range, not a reassurance.

• “Can I have the method document, in writing, before we sign?” Reproducibility and documented method are two of the eight decision-grade criteria. A vendor that cannot produce a document has already answered question 9 for you.

Final thought... We are taking bets that by the end of 2027, tier language will appear as a scored line item in AI visibility vendor selection RFPs, and the vendors who cannot answer question 9 in writing will stop making shortlists. If that has not happened by December 2027, this post was wrong. What do you think?