What is share of answer — and how do you run your first baseline?
“Are we showing up in AI?” is a feeling until you count it. Share of answer is how we count it — our own metric, and the one number we check before and after everything we publish.
AI visibility
“'Are we in AI answers?' is a feeling. Share of answer is the number that ends the argument.”
The answer in 30 seconds
- →Share of answer is how often an AI assistant names your brand across a fixed set of buyer questions. Our own metric, reported as mentions out of total.
- →To run a baseline: list 20–50 real buyer questions, run them across ChatGPT, Perplexity, Gemini and Google's AI Overviews, and count the answers that name you.
- →Freeze that number before you publish anything. It's worthless as a boast and priceless as a before-and-after.
- →Ours was 0 of 46 across four engines in July 2026. We publish it on purpose — the point is the second measurement.
What is share of answer?
Share of answer is our own measure of how often an AI assistant names your brand when someone asks a buying question in your category. We run a fixed set of real buyer questions across four engines and report the answers that mention you as a share of the total. It's deliberately plain — a fraction, not a score out of a hundred. Mentions over total. That's it.
How do you build the question set?
Start with the questions your buyers actually type, not the ones you wish they'd type. Pull them from sales calls, support tickets and the “people also ask” boxes under your category. Twenty to fifty is plenty. Keep them fixed — the same list every month — because a baseline you keep changing isn't a baseline, it's a mood.
[ figure — the fixed question list, the four engines beside it, and last month's count next to this month's ]
How do you run the first baseline?
Ask each question in each engine, in a clean session, and record whether the answer names you. Do it across all four — ChatGPT, Perplexity, Gemini and Google's AI Overviews — because visibility in one says nothing about the others. Add up the mentions, divide by the total, write the date on it. That frozen fraction is your zero. Everything after is measured against it.
What's a good number — and what do you do with it?
There's no universal “good”; there's only up from your own baseline. Zero is fine to start — ours was. What matters is the trend: which questions you win first, which engine warms up, which competitor holds the seat you want. Re-run monthly, because assistants favour fresh sources and the picture shifts.
Questions your team will ask
Why “share of answer” and not “share of voice”?
Share of voice counts presence in media you can buy or place. Share of answer counts whether an independent assistant chooses to name you. You don't control the answer, which is why the number is honest.
Do I need a paid tool?
No. You can run a first baseline by hand in an afternoon. Tools help you repeat it at scale, but the method is free.
How many questions and engines?
Twenty to fifty questions, four engines. Fewer questions and the number gets noisy; skip engines and you miss where you're absent.
How often should I re-measure?
Monthly. Assistants refresh sources often, so a quarterly check hides most of the movement.
AI visibility