A citation gap is a source that AI assistants repeatedly cite when answering your category's buyer questions — but that never mentions your brand. It's the single most actionable finding an AI-visibility scan produces, because the engine has already told you it trusts that source. You don't need to earn the engine's trust; you need to show up where it already reads.
1 — Why citation gaps beat every other fix
Getting recommended by AI has a supply chain: engines read sources, sources name brands, answers repeat what sources say. Most "AI SEO" advice attacks the wrong end — rewriting your homepage for machines the engine rarely reads. A citation gap attacks the supply chain where it's already flowing: the page is being read today, in your category, for your questions. The only missing part is you.
2 — What the gaps look like in measured data
Every category we've studied has a concentrated citation corpus — a handful of hosts feeding most answers:
| Study | Most-cited sources | The gap |
|---|---|---|
| SaaS payments | Rival docs & comparison pages out-cited official sites | Vendors absent from the comparison pages AI reads |
| B2B suppliers | Wikipedia cited 36× across 120 answers | No supplier had a presence in the encyclopedic layer |
| GEO industry | Listicles and stats aggregators | Real vendors missing from machine-written lists |
The pattern repeats: third-party pages out-cite official websites in every category studied so far. Your site is rarely the lever. The lever is the short list of pages the engines keep returning to.
3 — How to find yours in 20 minutes
Manual version, no tools: take 6–10 real buyer questions and ask them on Perplexity and Google AI Overviews — the two engines that cite most consistently (ChatGPT showed no sources on half our Chinese-language checks). Copy every cited URL into a sheet. Tally by host. Now cross out the hosts that already name you. What's left, ranked by citation count, is your gap list — and the top of that list is your warmest door.
4 — How to close a gap (and what doesn't work)
What works: getting genuinely listed where the source already lists others — a review-site profile, a directory entry, a pitch to the author of a comparison article that's missing you, a community answer written by a human with your name attached. Each is ordinary marketing work; the citation data just tells you which of it matters.
What doesn't: paying for "guaranteed AI placement". Nobody sells positions inside AI answers — engines don't take placement money for organic answers, and anyone guaranteeing a rank is guessing with your budget. The honest version of this work is measurable, not guaranteed: close a gap, re-scan weekly, watch whether the answer flips.
5 — Gaps differ by engine, so measure before you spend
Each engine reads a different corpus: in our studies the same category showed different top-cited hosts on Perplexity than on Google AI Overviews, and ChatGPT often cited nothing at all. Before investing in any source, check which engines actually read it for your questions — a gap on the engine your buyers use is worth ten on one they don't.
Where these numbers come from
Figures are from Ainswer's published multi-run studies: SaaS payments (235 scored answers), B2B supplier shortlists (120), who feeds the GEO answer (186), and retrieval & local brands (307). Raw verdicts included in each.
See your citation gaps, counted.
Paste your site — one real buyer question across five engines, free, no email. The full scan tallies every cited source in your category and flags the ones that don't name you.