2026-09-15
Share of answer: the KPI that will replace your ranking report
For twenty years, "how visible are we?" meant a ranking report: your position for a keyword list. That made sense when a search result was ten blue links and position #3 still got clicks.
An AI answer is not ten links. It's one synthesized recommendation, usually naming two to four brands. You're either in it or you don't exist for that customer.
Defining the metric
Share of answer = of the buying-intent questions in your category, the percentage of AI assistant answers that mention your brand or store.
Measured properly, it has three dimensions:
- Coverage — which of ChatGPT, Claude, Gemini and Perplexity mention you at all. They have different crawlers, different data partnerships, different blind spots.
- Position — first recommendation or a trailing "also consider".
- Trend — answers change as models update and competitors optimize. A one-off check is a snapshot; the value is in the time series.
Why it moves
Share of answer is downstream of boring, fixable infrastructure: whether AI crawlers can read you, whether your products expose structured data, whether your content is quotable. In our audits, stores that go from blocked-and-unstructured to fully AI-readable typically see their share of answer move within one to two model-update cycles.
That's also why it's a fair KPI: unlike rankings, you're not fighting an opaque algorithm with content volume. You're mostly fixing the machine-readability of what you already have.
Start measuring
Define 10–50 questions your buyers actually ask ("best X under €Y in Spain"), run them across all four assistants weekly, and track the mention rate. That's exactly what the aiListing platform automates — starting with a free audit that shows whether AI systems can read your store at all.