Why AI assistants cite some sites and not others

Every engine that searches the web before answering has to pick a few pages to read. Those pages become the citations under the answer, and the brands named on those pages become the brands named in the answer. Getting mentioned by an AI assistant is mostly getting mentioned on the pages it reads.

Across thousands of collected answers the same kinds of sites come up: comparison and review publishers, community threads on Reddit and similar forums, review platforms and directories, retailer and marketplace listings, and Wikipedia for background. Your own site is cited too, but mostly for questions about you, rarely for questions about the category.

Reachability differs by site type. A review platform lets you claim a profile today. A directory takes a listing. A community answers to honest participation, not pitches. A publisher's roundup is reachable when the editor updates it, which is a pitch with something real to offer: a product to test, data, a quote. Wikipedia and government sites are not reachable by outreach at all.

This is why per-prompt citation lists are so useful. For each question you want to win, the engines tell you exactly which pages they trust. Read those pages, see who is named and why, and pick the two or three where a mention is realistic.

On your own site the work is different: answer the exact question on one page with clear entity names, comparable facts and dates, keep it crawlable for the AI user agents, and describe it in structured data. That gives an engine something it can cite when the community pages fall short.