Sources

See which domains and pages AI cited in its latest answers, and organise their source categories.

Sources overview

Open AI visibility → Sources. The header shows the latest scan information and the counts of domains, answers with sources and answers. Category, model and funnel stage filters narrow the results.

Sources uses the latest completed answers for each prompt and model, excluding prompts that name a brand. A source is a page the answer cited. Search results an AI model looked at but did not use in its answer are not counted. Counts are answers: a domain cited several times in one answer counts once for that answer.

Use Source trends for changes across scans and for pinned external URLs.

Read the source mix

Source mix shows the share of citations by category, each domain counted once per answer. Click a category in the legend to hide or include it. Percentages are recalculated across the categories still visible in that mix; this legend does not filter the domain tables below.

The web-search indicator reports whether answers were recorded as using web search. An answer with no cited URLs is not, by itself, proof that no web search occurred.

Open cited pages

Domains lists domains ordered by how many answers in the latest scan cite them. Use the category filter, the unrecognized-only control when available, and pagination to find a domain.

When you have queries in Rankings, the In Google results column shows on how many of those queries the domain appears in Google results, and Only in Google results keeps only those domains.

Click the answer count to expand the cited URLs. Open a URL to inspect that page. Clicking the domain name instead opens the domain’s website.

An arrow beside an answer count compares it with the previous completed scan available for the relevant prompts and models. The first scan has no previous value to compare.

Set a source category

Owners, admins and members can choose a category from a domain row; viewers can read the classification. Use Unrecognized — set a category to identify a domain still awaiting classification, then select the category that describes the site.

Your own domains and tracked competitor domains have classifications derived from their registered websites. These rows have a fixed label rather than an editable category picker. Review Brand or Competitors if a website is attributed to the wrong brand.

Inspect the sources of each brand

Competitor sources shows, for your brand and each tracked competitor, the domains that are its sources. A page counts as a brand’s source only when an answer that names the brand cites it and the page itself carries the brand. Expand a count to see the URLs.

The count next to a brand is the number of answers. The addition {count} next to the name counts strong sources: answers in which the citation stands in the same part of the answer as the brand’s name. Cited pages are checked for brand names after the scan; pages still being checked appear later, and pages that could not be checked are not counted.

A domain listed for a competitor is evidence for the pages that were cited. It does not establish that your brand is absent from every page on that domain. Open the relevant pages before treating the difference as a content gap.

For a specific answer’s sources, open Prompts. For citations to your own pages alongside visits and search data, use Traffic & results.