Working Citation Gaps

A citation gap is a prompt where your site should be cited based on relevance and volume — but isn't.

Where to find them

On the Citations page, open the Prompts tab and select the Not cited filter. The list sorts by winnability, easiest wins first, using what ChatGPT searched for (below) and how authoritative the sites cited instead are. The domains cited instead are shown beside each prompt. The same gaps, rolled up by page, also appear on Opportunities as AI Citation Gaps.

Why gaps matter

Each gap is a concrete, demand-backed opportunity to create or improve content that AI engines will cite.

Analyze, then send to optimization

Expand a not-cited prompt and click Analyze & Recommend to compare your page against the top competitor that got cited (client admins only — each analysis has a small per-run cost). The recommendations can then be sent straight into your normal pipeline with Send to Optimization. See Submitting a Page.

What ChatGPT searched for

Before it answers, ChatGPT runs a few web searches. Expand a prompt to see them under Searches ChatGPT ran, with how often each one recurred. The badge under a not-cited prompt's status sums them up:

• Searches for you: it searches for your brand by name and still cites someone else. Fix these first; it's usually a page or indexing problem.
• Open search: it searches the topic itself. An open contest, and the most winnable kind.
• Wants official sources: it asks for official or research-backed sources without naming one. Winnable by pages that read as authoritative: data, named studies, expert review.
• Comparison: best-of or versus searches that name you alongside competitors. Review sites tend to win these.
• Wants a competitor / Wants an institution: most searches name a specific competitor or institution. Usually not worth chasing.

An asterisk (*) after a badge means it's provisional: it's based on only a few runs so far, so it counts for less in the sort. Searches are recorded from mid-September 2026 onward, so prompts from earlier scans have no badge.

When you run Analyze & Recommend, the analysis checks your page against those searches too, and each recommendation lists the searches it answers. Those show up as the target keywords when you send it to optimization.

Prioritizing

Start with Searches for you, then Open search gaps. Skip gaps where the competition is Institutional (.gov, major medical centers) or where ChatGPT wants a specific competitor or institution; AI defers to those. Among the rest, chase the Peer- and Low-tier gaps first; those are proven, winnable topics.

Custom prompts and source pages

Custom prompts you add yourself start with no linked source page, so the analyzer has nothing to compare against the competitors. The gap panel un-blocks this two ways:

• Previously cited — if one of your pages won this prompt in a past scan, the panel lists it with the exact citation evidence. Click Attach & Analyze to link that page and run the comparison in one step.
• Find a page for this prompt — for prompts that were never cited, this button looks at which of your pages already rank in Google for related queries and recommends the best fit (client admins only; each run has a small cost). If a page fits, attach it with one click. If nothing on your site fits, you get a new-page brief — a suggested title, URL slug, and angle describing what the page must cover to win the citation. Publish the page, attach its URL, and the next scan tracks it.

Once a source page is attached, the prompt behaves like any panel prompt: gap analysis, Send to Optimization, and outcome tracking all work. A prompt's source page can be set once — delete and re-add the prompt to change it.