Sources tab

The Sources tab is the expanded version of the Sources block on the dashboard: every domain cited across your prompts, with the filters needed to turn the list into a plan.

The two leaderboards

Domain sources — one row per domain, with:

• Citations — times it was cited across answers in the period
• Citation share — its share of all citations counted
• Coverage — the share of answers citing it at least once
• Domain Rating — a 0–100 estimate of the domain's link authority
• Type — what kind of source it is: Owned, Competitor, UGC, Editorial, Reference, Institutional, Corporate or Other

Page paths — the individual URLs behind those domains. This is where the leaderboard becomes a brief: knowing that a review site dominates your topic is useful, knowing it is one "best X for Y" article doing the work is actionable.

Column definitions and how to read the gap between citations and coverage are on Citations & Sources.

Filters

Group. Pick the prompt group to analyse — one at a time. Sources are highly topic-specific: the domain that dominates your comparison prompts is often absent from your how-to prompts, which is exactly why the tab is scoped this way rather than pooling everything.

Assistant. Assistants draw on different sources. Filtering to one shows the source landscape that assistant actually uses.

Period. Compare shares rather than raw counts between periods; a longer period simply contains more citations.

Source coverage by topic

Above the leaderboards sits a single row that answers a different question: for the selected group and period, how many answers were analysed, and how they split between

• With sources — answers that included at least one visible citation, and
• Without sources — answers with no visible source, produced from the model's own knowledge rather than from pages it fetched.

That split sets the ceiling on what the leaderboards can tell you. If most answers in a topic come without sources, the citation work below has less leverage there — you are up against what the model already believes about your category, which moves slowly and through breadth of coverage rather than through any single page. A topic that is mostly with sources is the opposite: the pages listed below are the live inputs, and changing them changes the answers.

Turning the list into work

1. Sort by coverage, not citations. Coverage is what fraction of answers pass through the source.
2. Check the type. Each calls for different work: your own pages you can edit directly, competitor pages tell you what you are up against, and third-party sources are where accuracy about you matters most.
3. Read the source as the assistant does. Open the page and check whether your brand appears, and whether what it says is current and accurate.
4. Work in this order — your own pages first, since they are the fastest to change; then third-party pages that describe you inaccurately or out of date; then the sources where you are missing altogether.

Things worth knowing

• Not every answer exposes sources. Answers without citations still count toward visibility metrics but contribute nothing here.
• Your own domain ranking first is good but fragile. Assistants weight third-party corroboration; a leaderboard made only of your pages tends not to hold.
• Domain Rating is context, not a goal. A niche site with high coverage in your topic beats a high-authority site nobody cites for these questions.

Related

• Citations & Sources — the metric definitions
• Prompt gaps — pair a gap with the sources behind it
• Schema audit — making your own pages easier to use as a source