> ## Documentation Index
> Fetch the complete documentation index at: https://docs.surnex.io/llms.txt
> Use this file to discover all available pages before exploring further.

# ChatGPT visibility

> See what ChatGPT says for a query and which sources it cites.

**AI → ChatGPT Visibility** runs a query through ChatGPT and shows the response together with the sources behind it.

## Running a lookup

Enter a query in **Enter a keyword to search ChatGPT…** and submit. You get the **ChatGPT Response** and a table of sources. Past lookups are kept so you can revisit one without re-running it.

## The sources table

| Column      | What it is                         |
| ----------- | ---------------------------------- |
| **Source**  | The cited page                     |
| **Domain**  | The domain it's on                 |
| **Snippet** | The passage drawn from it          |
| **Scores**  | Relevance scoring for the citation |

**Snippet** is the most useful column on the page. It shows the exact text ChatGPT extracted — so when your page is cited, you can see *which* passage earned it, and when a competitor is cited, you can see what they said that you didn't.

**Domain** repeated across queries identifies the sources ChatGPT treats as authoritative in your market. Those are outreach targets: being mentioned on a source the model already trusts is often more effective than optimizing your own page.

## Why this differs from Google

ChatGPT and Google's AI features answer from different foundations, so results diverge in ways worth expecting:

* **Different source preferences.** ChatGPT frequently favours documentation, community discussion, and long-form comparisons over pages that rank well organically.
* **Ranking doesn't transfer.** Position one on Google grants nothing here. Sites that rank modestly are cited regularly, and vice versa.
* **Answers vary between runs.** The same query can produce different phrasing and a different source mix. Treat one lookup as a sample, not a measurement — run important queries more than once before concluding anything.

That last point is the practical caveat. A single absence isn't proof you're never cited.

## What to do with it

Run your highest-intent commercial queries — the ones where someone is choosing a vendor — rather than informational terms. Those are where an AI answer displaces a decision you'd otherwise have influenced.

For each:

1. Are you mentioned at all?
2. If not, which sources were cited instead?
3. Do those sources cover you? If a comparison site is cited and doesn't list you, that's the gap — and it's fixed by getting listed, not by editing your site.

[Citation gap](/ai/citation-gap) does this comparison systematically across several keywords and competitors at once.

## Empty response

**No response content available.** The lookup returned nothing usable — retry, and if it persists, the query may be one ChatGPT declines to answer substantively.

## Usage

Each lookup draws on your plan's allowance. See [Usage and limits](/billing/usage).