What AI Citation means
Citations appear differently across platforms — Perplexity shows numbered source cards, ChatGPT sometimes links inline, AI Overviews shows source chips. All represent the same thing: attribution.
A citation requires the source to have been retrieved, selected, and used. Each stage eliminates candidates, which is why being crawlable is necessary but far from sufficient.
Why it matters for AI search
Citations are the unit of AI visibility. They put your brand in front of a buyer at the moment of research, whether or not a click follows.
They also carry implied endorsement in a way a search result doesn't — the model chose you as a trustworthy source rather than listing you among ten options.
See it in action
From candidate to citation, and what drops out at each stage.
Seven of a hundred made it.
Most of the loss happens between retrieval and extraction, which is a structural problem rather than a content-quality one. That's usually where the cheapest wins are.
How to get it right
Earning AI citations
- Be retrievable — crawler access and rendering come first
- Make passages self-contained so extraction doesn't distort them
- State specifics that are checkable rather than generic claims
- Build corroboration so your claims aren't single-sourced
- Track citations against a fixed prompt set rather than by screenshot
Common questions
Do AI citations drive traffic?
Sometimes, less than a search result would. Their value is partly visibility at the research moment, whether or not a click follows.
Can we buy AI citations?
No. There's no placement mechanism. Anyone offering guaranteed citations is misrepresenting how these systems work.
How do we know if we're being cited?
Run a fixed prompt set across platforms on a schedule and record it. Perplexity is easiest since it shows its source list openly.
Related concepts
These come up alongside AI Citation constantly.