Learning

How to measure AI visibility: rates, not scores

A practical method for reporting AI visibility with defined rates, denominators, separate question layers and assistant types.

· 5 sections

“How do you measure AI visibility?” A dashboard might offer a score, position or grade. Those summaries are difficult to assess when the metric’s inputs and calculation are not explained. In our method, we report defined event rates with their denominators and retain the responses behind the results so the reader can inspect what was counted.

This guide presents one auditable format: rates, or counts of a defined event over a defined population, printed with the denominator. The method is designed to make the outcomes we report inspectable.

Read it once and you can run the measurement yourself: four rates, three layers, three assistant types, each reported separately with its denominator.

Where Business Hub fits

OpenAI’s Business Hub adds aggregated ChatGPT activity to the evidence you can review. Platform activity and a controlled answer study serve different purposes: one reports activity recorded by ChatGPT, while the other examines responses to a selected question bank. Keep those measures separate. The Business Hub guide for Mexican businesses explains the setup process and the information you can prepare now.

The four rates

Ask a bank of real buyer questions to the assistants the study covers, several times each, and read every response. For each brand, count four things:

  1. Mentioned. The brand appears in an answer to a question that never named it. The floor: presence in the category conversation.
  2. Recommended. The assistant actively suggests it. Being listed among eight is not being told to buy it.
  3. Named as the best. The answer distinguishes it instead of grouping it. The strongest and rarest form of presence.
  4. With a purchase path. The answer says where to buy it. The closest honest proxy to purchase intent, with no analytics needed.

Each one is a rate with its denominator. From our invented sample, a sunscreen brand measured against Bloqsol and Dermalux: Mentioned 26%, 103 of 396 responses. Recommended 11%, 44 of 396. Named as the best 3%, 12 of 396. With a purchase path 7%, 28 of 396.

A percentage with no population behind it is a decoration. 26% of what? Of 396 responses, each one read. Now you can audit it.

The three layers, reported separately

The bank has three layers, and they answer different questions:

  • Unbranded questions never name the brand. They measure spontaneous presence, the only honest way to know if you appear on your own.
  • Brand-named questions ask about you directly: what the model believes, where it sends buyers, which critique arrives attached.
  • Competitive questions map who owns the conversation, head-to-head and in comparisons that happen without you.

Our method reports each layer separately because each answers a different question. We do not average them into a single figure, which could obscure those differences.

The three assistant types

The same brand measures differently depending on how the assistant works, so the types are reported separately:

Assistant typeMention rate (invented sample)
Memory-only call9%, 12 of 132 responses
Search-then-answer30%, 39 of 132 responses
Google AI Mode39%, 52 of 132 responses

In this invented example, the search-enabled groups have higher mention rates than the memory-only group. The figures illustrate the reporting format; they do not establish why the rates differ or predict results for other brands. For source metrics, if an assistant does not surface sources, record that as unavailable rather than as zero cited sources.

Why this beats a score

Three practical consequences:

  • You can audit it. Every figure traces to saved responses you can read.
  • You can reproduce it. The same frozen bank re-run gives comparable numbers, so trends mean something.
  • You can act on it. “Recommended is low, search-then-answer is carrying us, the retailer page is the top cited source” is a plan. “Our score went from 61 to 64” is a shrug.

Everything above is the method in short form; the full version, including how the bank is built and what the study cannot tell you, lives here: /metodologia/. To have the method applied to your category, see the AI visibility study.