Skip to main content
Optimizer.team
Start free
AEO, GEO & LLM core

Citation vs Recommendation

A citation is when an AI links to your page as a source; a recommendation is when an AI names your business as the answer. A third outcome, retrieval, means your page was read but never used. All three matter, and they reward slightly different content.

What it is

In AI search, one outcome gets flattened into three different things: retrieval, citation, and recommendation. They feel similar but they're produced by different parts of the system and rewarded by different content.

Retrieved means an AI read your page and listed it as a source — in Perplexity's Sources panel or Links tab, for example — without attaching it to any specific claim in the answer. It shows the system found you; it doesn't mean the answer used you.

A citation is when an AI tool quotes or summarizes a specific claim and links it back to your page as evidence. Perplexity does this constantly, showing the source as a named chip next to the sentence it supports. Google's AI Overviews cite multiple sources in a sidebar. ChatGPT with web search cites pages it relied on. Citations tend to reward reference-style writing — clear claims, specific numbers, dates, structured sections.

A recommendation is when an AI names your business as the answer — "you should hire X" or "the top option is Y." Recommendations are common in assistant-first conversations ("who's the best plumber near me?"), but research-first tools recommend too: in testing, Perplexity has named specific local businesses as its pick, drawing the supporting facts from the businesses' own sites, Yelp, and directories. Recommendations tend to reward entity clarity, strong reviews, and consistent public signals — anything that makes the system feel safe naming you specifically.

Why it matters for your business

If you only chase rankings, you may miss all three. Retrieved-but-never-cited means the system found your page but didn't quote it — usually a sign the answer-first claim isn't sharp enough. If you only chase citations, you'll generate referral clicks but the user still has to pick someone — and the AI might pick a competitor with cleaner local signals. If you only chase recommendations, you might miss the high-intent research traffic that comes from being a trusted source. The win is tracking all three and writing content that earns each one.

Example

A pest-control company wrote a "what's normal pest activity in [city] by season" guide with month-by-month data. That guide gets cited by Perplexity and Google AI Overviews — earning referrer traffic. Meanwhile, their core service pages, GBP profile, and review responses are tuned for entity clarity and proof — earning recommendations in ChatGPT, and in Perplexity too, when locals ask "who should I call for ants in [neighborhood]?" The same business earns different outcomes from different pages, on the same surface and across surfaces.

Related

Related