> ## 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.

# Citation gap

> Find keywords where competitors are cited in AI answers and you are not.

**AI → Citation Gap** compares your domain against competitors across a set of keywords and shows where they're cited in AI answers and you aren't. It's the systematic version of running [ChatGPT visibility](/ai/chatgpt-visibility) one query at a time.

## Setting up a comparison

<Steps>
  <Step title="Add competitor domains">
    Type domains into the input rows — `competitor.com` — up to **10**. **Add** creates another row; the remove control drops one, and at least one always remains.

    **Load saved competitors** fills the rows from your project's [Competitors](/competitors/add) list, up to the first 10. Faster than retyping, and it keeps this analysis consistent with the rest of the project.
  </Step>

  <Step title="Enter keywords">
    The **Keywords** box takes up to **50**, separated by commas or line breaks. Over the limit you'll see *"Maximum 50 keywords per request"* and nothing is submitted.
  </Step>

  <Step title="Run the comparison">
    Results come back as a grid of keywords against domains.
  </Step>
</Steps>

## Reading the results

Each keyword row shows which domains were cited in the AI answer for it, with your domain first.

A row where competitors are cited and you aren't is tagged **GAP** in red. Those are the actionable rows — the query has an AI answer, that answer names sources, and you're not among them.

Rows also show what you were **Co-cited with** where you do appear: the other sources sharing the answer with you.

## Prioritizing gaps

Not every gap is worth chasing. Rank them:

**Gaps where several competitors are cited.** If three of your competitors appear and you don't, the answer clearly has room for vendors like you and you've been left out. That's the strongest signal in the report.

**Gaps on commercial queries.** A gap on a term where someone is choosing a vendor costs more than one on a definition. Cross-reference [CPC and intent](/keywords/metrics) from keyword research.

**Gaps where one competitor is cited.** Weaker. Could be a genuine authority advantage, could be variance in a single run.

Deprioritize rows where nobody is cited — that query's AI answer isn't drawing on vendor sources at all, and there's nothing to win.

## Co-citation

The **Co-cited with** detail is the underused part of this page. It shows the company you keep in AI answers.

Two things to look for. If you're consistently co-cited with the same publications or comparison sites, those are the sources shaping answers in your market — being covered by them is leverage. And if a competitor is co-cited with you nearly everywhere, they're your genuine peer in how these systems see the market, regardless of how you compare on rankings.

## Choosing keywords

Fifty is enough to cover a real question set. Spend them on:

* **Commercial and comparison queries** — "best X", "X vs Y", "X for \[use case]". This is where AI answers displace a decision.
* **Conversational phrasings** — how someone would ask an assistant, not how they'd type into a search box.
* **Queries you already rank for** — a gap there is the sharpest finding: you have the organic authority and still aren't in the answer.

Avoid navigational and brand queries. You'll be cited for your own name, which tells you nothing.

## Variance

AI responses vary between runs, so a single gap isn't proof. Re-run the important keywords before acting — a gap that persists across runs is real; one that appears once may not be.

## Related

* [ChatGPT visibility](/ai/chatgpt-visibility) — the full response and source snippets for one query
* [LLM benchmarking](/ai/benchmarking) — whether the gap exists on every platform or just one
* [Backlink gap](/backlinks/gap) — the same idea applied to links