Analyze AI visibility
Sources
See which domains and pages AI cites for your prompts so you can close citation gaps and shape what models treat as evidence.
View Markdown · AI index · Full documentation for AI
Overview
Sources shows which domains and URLs appear when AI models cite the web in answers to your monitored prompts. It complements brand metrics — visibility, share of voice, and sentiment — by revealing the evidence stack behind those answers: who earns citations, where you show up, and what to publish or pitch next. Use Sources alongside Pages when you tie citations back to owned URLs.
New to the workflow? Read Step 4: Understanding your sources in the Quickstart Guide.
Why sources matter
Models lean on cited pages when they explain categories, compare options, or list “best” picks. Knowing which domains dominate citations in your slice tells you where narrative and SEO effort pay off — and where competitors earn airtime you do not.
- Citation benchmarks: See which sites models treat as authoritative for your prompt mix.
- Gaps: Spot domains rivals earn repeatedly versus your properties or industry coverage.
- Action: Prioritize content formats, partnerships, and technical fixes that move the URLs you care about into the citation set — often a stronger lever than tweaking a single prompt line.
Where to open Sources
In the sidebar under Monitoring, open Sources (/sources). For the full sidebar map, see Sidebar navigation.
Filters and time range
Sources uses the same workspace filters and date / time-range controls as the rest of the dashboard (aligned with how you scope prompts and reports). Change the slice here and the domains, counts, and chart update to match. Filter and date controls align with other monitoring views — see Sidebar navigation for where those live.
Top of the page: score card and trend
The top row pairs a score card with a trend chart. The card highlights total citations and how they break down. On the all-domains view (/sources), the breakdown uses domain types from the product — Reference, Corporate, UGC, Editorial, Institutional, and Other — so you can see whether models lean on encyclopedic sites, brands, forums, publishers, or other buckets for your current scope.
The trend chart summarizes citation activity over time for that same scope. When you drill into a single domain, both widgets reflect that narrower slice.
Domain list
The main table lists domains ranked for your filters, with pagination so you can move through large result sets. Select a domain to open its detail view.
Open a domain
Choosing a domain navigates to /sources/<domain> (the domain is URL-encoded in the path). That screen keeps the score card and trend focused on that host while the table shows the specific URLs cited under it — again with pagination and the search and sort controls shipped in the product.
URL grouping in the UI
You may see labels that group URLs (for example article versus listicle-style paths). Those groupings come from lightweight URL heuristics in the interface to help you scan long lists — they are a convenience for reading the table, not a guarantee of a fixed editorial taxonomy on the backend.
When counts still look noisy
Give prompts time to accumulate before treating Sources as definitive — early days can swing as runs refresh and your prompt mix stabilizes. Pair this view with prompt quality and Competitors so citation patterns match the questions you really care about.