What We Disclose — Item by Item

What we disclose, item by item

Before anyone buys a measurement, a few things have to be checkable. Where the questions went, how the question set was built, how the readings were validated. We have filled in our answer for each of the 13 items below, and every row links to the page where that value actually lives.

Disclosure item	What we publish	Where
Platform Coverage	Every place we ask is named on the record, model version included. The roster differs by market and the US one is published in full.	Open
Prompt Library Construction	How many questions go out in a round, and how many times each one is repeated. The thresholds a round must clear are on the same page.	Open
Query Sourcing	Whether a query is synthetic or drawn from real search demand is stated, along with the selection procedure behind the set.	Open
Source Type Classification	Sources an answer pulls in are grouped by type, and the rule that puts a source in one group rather than another is published.	Open
Data Collection Architecture	Whether readings come from active querying, panel observation, platform-native data or a mix of those is stated up front.	Open
Data Collection Methodology	We check our own collection against each platform's terms, rate limits and access rules first. Methods that failed that check were withdrawn rather than kept quiet.	Open
Panel Validity	The share rule that puts a place on the roster is published, and the same rule is applied in every market we sell into.	Open
Data Validation	How many readings we failed to take, and where our own instruments were wrong, is republished every round in the same place.	Open
Mention Logic	A mention is the brand name appearing in the answer. A separate check keeps a different company with the same name out of the count.	Open
Attribution Logic	A citation is the answer carrying the domain as a source. That is a different number from a bare mention and the two are never merged.	Open
Sentiment Logic	The rule that reads how an answer speaks about a brand is written in the glossary, so the word means the same thing on every page.	Open
Factual Accuracy Classification	Statements published on our surfaces pass verification first, then go into a public register with their source type and reference date.	Open
Historical Data and Versioning	When a baseline is reset, the date and the reason are recorded, along with how readings either side of it should be compared.	Open

Why a table

More than twenty companies now sell AI visibility measurement, and the same brand comes back with different answers depending on who you ask. Buyers have budget ready and no basis for comparing what they are buying. So we fill in our own column. Everything here points to a page that was already public; this table is the index that puts those doors in one place.

Consider what the empty column costs. Without it, two proposals sit side by side and you cannot tell which one asks more questions in more places, so the decision drifts to whichever quote reads better or costs less. What you did not buy surfaces when the first report lands. A quarter is gone by then, and the budget is already committed.

The scale is public as well. A panorama asks 50 questions, 7 times each, across 15 engines. It counts 17 surfaces where those answers actually show up. An engine is one service that produces an answer; a surface is a screen a person actually looks at.

Why us, in one paragraph

Every reading is sealed with the conditions it was taken under, which is why a figure you receive can be checked against a record instead of taken on our word. Readings the grid planned but could not take stay in the denominator and are counted next to the ones that read cleanly. Dropping them would improve the ratio on its own; we do not make that calculation. How many competing brands ride along in the same grid is written into the contract as a number, tier by tier. Every row in the table above opens a page that was already public, for the same reason.

How to read this table

The item list comes from an industry framework published in August 2026. That document states plainly that it does not rate measurement providers, and we make no claim of certification or compliance with it. We are doing one thing here: writing our answer next to each item it asks providers to disclose.

Source: IAB AI Visibility Measurement Framework (2026-08) — https://www.iab.com/wp-content/uploads/2026/07/IAB_Measuring_Visibility_in_the_AI_Era_August_2026-1.pdf

You can read how we measure before you decide anything.

We agree the question set and the roster with you, measure exactly as described above, and hand over what came back.

Ask about an audit