AI Citation

A citation puts a business in front of a buyer while the buyer is still forming a view, and it carries a hint of endorsement because the engine chose that source. Even without a click, being named on a shortlist shapes what the buyer does next. The opposite risk is quiet: a company can be dropped from every answer in its market and see no obvious change in its traffic reports.

Engines also behave differently. Perplexity shows sources as part of every answer. ChatGPT cites mainly when it searches the web for that question. Google's AI Overviews link only a few pages. Being absent in one engine says little about the others.

When someone asks Perplexity, ChatGPT with search switched on, or Google a question, the tool composes one answer and usually lists the pages it drew from. Each listed source is a citation. A mention is different: the brand is named in the text but no page is linked. Both matter, but only a citation sends a reader somewhere.

Answers change from one run to the next, so a single test proves little. Counting only clicks hides citations that influence buyers without sending traffic. Chasing every engine at once, instead of the one your buyers actually use, spreads the effort thin.

• Write down 10 to 20 questions a buyer would really ask.
• Run the same questions in each engine once a month and record who is cited: you, a competitor or nobody.
• Rewrite the pages behind the gaps so the answer comes first and is easy to quote.
• Add original facts, such as your own numbers or a clear definition, since engines prefer sources that say something specific.
• Get the same claims repeated on independent sites, because a model trusts consensus more than a company's own page.