Module 1 tells you that you're missing from AI answers. Module 2 tells you why — and hands you a sequenced list of what to do about it. It runs on the citations Module 1 already collected.
The loop
Citations in
Every URL each engine cited for your prompts, already stored by Module 1.
Scrape the cited pages
Fetch each public source page — G2, blogs, Reddit threads, docs, news, Wikipedia — via the managed scraping API, plus your own and competitors' relevant pages.
Chunk + embed
Break each page into passages and embed them into pgvector, so passages can be compared semantically against the (sub-)queries they were cited for.
Score the signals
Score each page on the signals that drive citation — on-page extractability and structure, specificity, freshness, query match, and off-site authority. See Citability signature.
Learn the signature
For each query cluster, compare cited vs not-cited sources that appeared for the same query — which features separate the winners, and where your own content falls short.
Gap report + playbook
Render the source gap ("the engines cite these 8 sources; you're present in 2") and a prioritized playbook, sequenced by estimated impact.
The gap table
Per engine and merged across all three, the gap table lists the sources the engines cite for your category and marks which ones already mention you. The absent rows are your targets — the pages and profiles you need to get into or earn a mention on.
The prioritized playbook
The playbook turns the gap into sequenced actions, impact-ranked, with each one labeled as a selectability or discoverability lever — for example: "1. get listed and reviewed on G2 (discoverability, high impact); 2. add a 40–60 word answer block to this comparison page (selectability); 3. earn a mention in these 3 publications."
The signature drifts as the engines update, so the whole loop re-runs each cycle. For the conceptual split behind the two lever types, read Discoverability vs selectability.