Getting cited by AI

No. It cites sources when it browses the live web, which it does for some questions and not others. Answers drawn purely from training data usually name no sources, so brand presence still matters independently.

It helps your content be understood and indexed, which supports retrieval. Controlled testing found formatting-only differences had little effect on which retrieved source gets cited, so schema is worth doing but should not be your citation strategy.

Narrow a page to one question and answer it completely in the opening passage, with concrete figures and a visible recent date. Relevance and explicit specifics carry more measured weight than any presentational change.

On-page changes can affect browsed answers within weeks once recrawled. Building the recognition that gets you into the candidate set consistently takes months, because it depends on accumulated references rather than edits you control.

Yes, particularly on narrow questions where few credible pages compete. Small sites struggle on broad commercial questions where established sources dominate the candidate set, so specificity is the practical route in.

Usually because your page is broadly relevant rather than precisely relevant, or because the answer sits too far down to be extracted cleanly. Ranking gets you retrieved; being quoted depends on relevance and quotability.

Run a per-prompt diagnosis: list the queries you should win, check who is cited today, and compare your page against the cited sources at each stage of the citation pipeline. The earliest stage where the cited source clearly did better is your failure mode and your fix.

The page is never retrieved, is retrieved but buries its answer, is retrieved but not trusted enough, or is retrieved but answers a slightly different question than the one asked. Each is a different failure at a different stage and needs a different repair, which is exactly why diagnosing the stage first matters more than reaching for a standard fix.