Sources and citations
The domains assistants read before answering, and what to do about them
Why sources are the actionable half
Visibility tells you where you stand. Sources tell you why, and they are the only part of this product you can act on directly.
An assistant composing an answer about your category reads a set of pages and repeats what they say. If the three articles it reads rank you fourth, you are fourth — and no amount of work on your own homepage changes that. Getting into those articles does.
What Scorra records
For every answer, the domains and individual pages the assistant cited, with how often each was used and for which questions. From that you get:
- Source breakdown — how citations split across categories and content types. Answers the question where do the models actually read about my category: review sites, listicles, forums, vendor documentation, press.
- Cited sources — the individual domains and pages, ranked by how much work they do.
- Prompt impact — which of your tracked questions pull from one specific source. Use it to size the prize before you spend a month chasing a placement.
Turning that into work
The usual sequence: find the source that feeds the most questions you lose, check whether you appear in it, and if not, decide whether to pitch it, buy a placement where that is possible, or write the page that outranks it.
The agent can do the first three steps of that for you and hand you a draft.