Citations: the sources AI reads
Which domains and pages AI engines cite when they answer about your category, what the citation gap is, and how to use both to get mentioned more often.
Updated:
Some engines answer from memory, others answer by searching the live web and citing what they found. Citations are the second kind, and they are the most direct lever you have: get onto a page an engine trusts, and you get named in the answers built from that page.
Where citations come from
Citations are captured from Perplexity and from grounded Gemini responses, which return their sources with the answer. Engines that answer purely from training data produce no citations, so a low citation count is not automatically a problem, it may just reflect which engines you track.
The main page
Four numbers at the top: average citations per run, total citations, own-domain citations, and the share of runs that included any citation at all in the period.
Below that:
- Citations by AI engine: which provider cites you most.
- Citations by prompt type: which intent category actually drives sources. Comparison questions usually cite more than brand questions.
- Top cited domains and top cited pages: the ranked lists. Click a row for its own detail view with share over time, the pages behind a domain, and the prompts that triggered the citation.
- Source types: company, blog, social, news, government, other. A category where you have no presence is a content gap.
- Web search usage by provider: how often each engine bothered to search at all.
New sources
Domains that started citing you in this period and had not before. This is where a listing, a guest post or a directory submission shows up as a measurable result, usually a few days after it goes live.
Citation gap
The most useful block on the page: sources that cite your competitors but never cite you. Each row is a concrete place where the engines are already looking for answers in your category, and where your product does not exist.
Work down that list. A review site, a comparison article, a community thread: getting covered there is a much shorter path to more AI mentions than writing another page on your own blog.
What to do with it
- Close the gaps: pitch, submit or get listed on the domains from the citation gap list.
- Reinforce what works: if one domain drives many citations, get a second, deeper page there.
- Keep your own pages citable: clear structure, direct answers, real facts. See the AI Visibility Playbook and Site Scan.
Related: Competitors, Listings, Metrics explained.
Did the content help you?