The short answer

Reverse AI search is the practice of starting from a domain and working backwards to the complete list of buyer queries that ChatGPT already cites it on (Gemini and Grok are on the roadmap). It is the inverse of asking “how visible is my brand?” Rather than returning a single 0–100 score, reverse AI search returns the underlying queries: the exact questions, their intent, the page the model cited, and which competitors appear alongside it. The mechanic is an inverted index — a continuously refreshed record of which domains get cited for which questions — queried by domain instead of by question. It is the Ahrefs Site Explorer pattern for the answer layer of search.

What is reverse AI search?

Most AI-visibility tools answer a score question: they sample a few prompts, count how often you appear, and hand back a number. A number tells you almost nothing actionable — not which questions you win, not which you lose, and not who is being recommended instead of you.

Reverse AI search flips the lookup. We continuously ask ChatGPT the questions real buyers ask, record exactly which domains get cited in each answer, and store that in a query–domain inverted index. When you enter a domain, you don’t get a verdict — you get the evidence: every query the index has seen the domain cited on, the search intent behind it, the page it cited, and the competitors named in the same answers.

Why a query list beats a visibility score

A score is a lagging summary. A query list is a worklist. With the underlying queries you can do three things a score can’t support:

  • Triage. Sort by intent and see the bottom-of-funnel, ready-to-buy questions first — the ones where being named actually moves revenue.
  • Find the gap. See the exact answers where a competitor is cited and you are not. That difference is your AI keyword gap.
  • Verify. Every row traces back to a real model response you can read — no black-box scoring to defend in a client meeting.

What does a reverse AI search result look like?

The output is a table — one row per query a domain is cited on. The sample below is an illustrative example (not live index data) showing the shape of the result: the literal question, its intent, the model that cited the domain, and the competitors cited in the same answers.

Illustrative example — reverse AI search result rows (not live index data)
QueryIntentModelCompetitors named
best project management tool for agenciesCommercialChatGPTAsana, Monday
notion vs your-brand for client workComparisonChatGPTNotion
how to track tasks across multiple clientsInformationalChatGPTTrello, ClickUp
your-brand pricing for small teamsTransactionalChatGPT(none)

Read it column by column the first time, then row by row after that — the full field-by-field walkthrough is in how to read your reverse AI search report.

How the AI query index works

Under the hood it is a classic inverted index, repurposed for AI answers. A background worker runs buyer questions through ChatGPT, extracts the domains each answer cites, and writes the question → domain links into a shared index. Querying it by domain — the reverse direction — yields your AI citation footprint. Read the full mechanic in reverse domain lookup for AI citations.

Mention, citation, and the competitor view

Being mentioned by name and being cited with a linked source are different signals, and Mentn records citations only; the distinction matters for what you can act on. We cover it in mention vs citation in AI. Just as important is who you share answers with — the competitor query overlap tells you which rivals the model treats as substitutes for you, and why ChatGPT recommends your competitor and not you turns that into a fix list.

The full Reverse AI Search cluster

Every guide in this pillar, in reading order:

Where to start

The fastest way to understand reverse AI search is to run it on a domain you know. The free Domain Check returns the real query list — not a score — for ChatGPT today (Gemini and Grok are on the roadmap). Try your own domain, then try a competitor’s and compare the answers side by side.

For the wider research context on how often these rows change and how concentrated citations are by category, see our State of AI Citations work, and compare the tooling landscape on the compare hub.

Frequently asked questions

Is reverse AI search the same as an AI visibility score?
No. A visibility score collapses everything into one number; reverse AI search returns the underlying queries instead — the exact buyer questions a model cites you on, with intent, model coverage and the competitors named alongside. The score is a summary you cannot act on; the query list is a worklist.
Which AI models does reverse AI search cover?
The index records answers from ChatGPT today; Gemini and Grok are on the roadmap. Because models retrieve and weight sources differently, a domain can be cited by one and ignored by another — which is why coverage is reported per model rather than as a single blended figure. See share of model.
How is this different from Ahrefs Brand Radar?
Both apply a familiar SEO mental model to AI, but reverse AI search is query-level and citation-level: it hands you the literal questions and the pages cited on each, not an aggregate mention count. The closest analogy is Site Explorer for AI.
Do I need to install anything to run it?
No. The free Domain Check reads the existing query–domain index backwards for any domain — no script, tag or signup required. You enter a domain and read the real answers it shows up in.