The Citation Sources page works differently now. "Mentioned" used to flag any URL where your brand name appeared in the text — now it means an AI actually named your brand in a query that cited that URL. It's a narrower signal, but a more honest one.
Source classification got a full rework. The Type filter now covers social profiles, video platforms, code repositories, academic publications, analyst sites, PR wires, and app marketplaces — so filtering by type gives you something meaningful to act on. Owned sources are auto-detected from your social handles and website, and competitor sources are pulled from your scan's competitor list automatically. Where the auto-detection is wrong, a pencil icon on each row lets you reclassify a domain or handle prefix to Owned, Partner, Competitor, or Ignored — and that override sticks across all future scans.
Partner Sites is a new section and filter for domains you've paid for placement on or have affiliate relationships with. You can now see exactly how often that content surfaces in AI responses, which is useful if you're trying to evaluate whether a partnership is actually moving the needle. The Opportunity count has also been tightened — it now excludes vendor docs, marketplaces, code repos, and competitors, so it reflects realistic PR and outreach targets rather than every unclassified domain.
All the summary cards, the top domains list, and the source table now share a single filter state. Changing a filter moves all the numbers at once. Previously they could disagree. The credits usage breakdown no longer shows raw internal identifiers — each credit source now has a readable label.