Skip to content
← DocumentationReading the results

Citations and sources

Which domains shape the answers in a category, and which of them you can influence.

Every captured answer's citations are extracted, normalised to a domain, and classified as owned by the brand, owned by a declared competitor, or third-party.

What the source table shows

  • Domain and how many times it was cited across the period.
  • Ownership — yours, a competitor's, or neither.
  • Which prompts it appeared on, so a source can be traced to the questions it influences.
  • Click-through to the raw answer that cited it.

The uncomfortable finding

Across the field, roughly 84% of AI citations are earned or third-party media rather than the brand's own site. Directory listings, comparison sites, review platforms, trade press and forums do most of the work. A client who reads their citation table and concludes “we need more blog posts” has read it backwards.

Source churn is normal

Day-to-day overlap of cited sources for identical prompts runs around 34–42%. A source that appears once and vanishes is not a signal. A source that appears in a majority of runs across a fortnight is.

Share of voice, mention rate and citation shareRaw answer captures
Next step

See it against your own client list

A working demo runs your prompts, in your market, on live engines — not a sandbox with seeded data. Bring one client brand and three competitors.