How does Profound track brand coverage across AI answer engines?
Profound tracks brand coverage by collecting answer responses and recording whether the brand appears, according to its documentation read on 29 October 2026. Its published definition makes the denominator part of the result. Our own field-service capture on 22 September illustrates why engine labels also matter: Housecall Pro was named in 21 of 30 ChatGPT answers and 22 of 30 Google AI Overview answers.
What we measured
Reported. Profound's Answer Engine Insights guide describes a visibility score using responses with the brand over responses containing at least one brand. Its share-of-voice definition instead uses total brand mentions as the denominator. Citations are tracked separately. These are the provider's published definitions, read on 29 October 2026.
Measured. The Housecall Pro figures come from the September release, using ten field-service questions and three runs on each named engine. Each engine contributed 30 valid answers in this category. These observations were collected for the Index, not through Profound.
The Index counts cannot establish how accurately Profound collects or classifies responses.
What this means for you
Reasoned. You need the score's base. If a report drops answers that contain no brands, it describes coverage among brand-containing responses rather than coverage among all valid answers. Neither label tells you that difference by itself. Read the definition.
You also need the engine breakdown. In the Housecall Pro example, the ChatGPT result and the Google AI Overview result describe separate sets of answers even though both bases happen to be 30. Pooling them would answer another question. Preserve the individual totals when your decision concerns one engine.
For your own export review, ask for a row that contains the question, engine, date, brand-match decision and answer receipt. Then ask where a response without any brand goes. That single row and its missing-brand counterpart are more useful than a comparison of two unexplained percentages.
Also asked as
How does Profound track brand coverage across AI answer engines?
Related
Next step
Use Free 10-question check (one engine) to inspect a defined set of buyer questions. No email is needed; runs are limited per network and per day, and by the site's daily allowance.
For one documented gap, Fix one answer, $490. Keep the question and source receipt with the requested change.
Claiming or correcting a vendor record is free and never depends on a purchase.
Every figure comes from the published method, v1.1.
Prepared by the Broadcastwell team.
Page owner: Priya Nair, Management Consultant. Corrections: hello@broadcastwell.com