中文Get an audit

RankX Search Intelligence

How does RankX Search Intelligence measure visibility in AI answers?

We ask AI assistants the questions your buyers ask, using the same wording, location and settings each time. We then grade every answer for whether your business is recommended, cited, mentioned or absent. AI answers change from run to run, so we report what we saw, with quotes and links.

Written by Vincent Peng, Co-Founder. Reviewed by Gerry Gu, Co-Founder. Published 28 October 2026.
This page covers how we measure AI answers. For how we check your site's readiness, see our site-readiness method across nine areas.

Which AI assistants do you check today?

ChatGPT, Gemini and Google AI Overviews. Claude and Perplexity are coming soon.
For what each plan includes, see pricing.

Which questions do you ask the AI assistants?

We agree a list of buyer questions with you and check each one against Google autocomplete for your country and language (Google AU for Australian businesses), so they reflect what people actually search. Your brand name is checked separately, so it doesn't inflate the results.

Why can the same question get different answers?

AI assistants generate each answer fresh, so the same question can produce a different answer each time, even when nothing has changed on your side. One answer is a single sample, so we report what we observed and never turn it into a share of the market.
  • Recommended: the answer suggests your business as an option.
  • Cited: one of your pages or profiles is linked as a source.
  • Mentioned: you're named without being recommended.
  • Absent: you don't appear.
We grade each answer at the highest level it reaches, then report how many answers landed at each level, for each question and each assistant, with quotes and links as evidence. A result might read "recommended 2, mentioned 1, absent 2". That shows the format only, not a set number of answers. See how the work is done.

Why don't you give one overall AI visibility score?

A single score hides which questions, which assistants and how many answers it's based on. We report each question and each assistant separately, and mark anything we didn't check as "Not assessed", so you can see exactly what was measured. AI-answer results are reported as counts per rung (recommended, cited, mentioned, absent), not as one overall score.

What can't this measure?

  • Positions inside AI answers: not measured, because answers vary run to run.
  • Revenue or leads caused by AI search: not measured. We don't forecast revenue or leads.

Can you guarantee AI placement?

No. We don't guarantee rankings or AI citations. AI assistants are run by third parties and their answers are probabilistic, so no one can promise what they will say.

What does it cost?

See our pricing for current prices.

Method details