How Perplexity answers
Perplexity runs a search for every question, fetches the pages on the results with its own crawler, and writes an answer with a citation after each claim. It has an index of its own and it also reads pages live at answer time, so a page changed this morning can be cited this afternoon.
What it cites is visible in every answer, which makes it the easiest engine to learn from. Ask it your buyers’ questions and the sources list tells you which pages it trusted. In stored sessions those are comparison lists, review sites, recent articles and, for any question with an opinion in it, Reddit and forum threads.
Condition one: can Perplexity read you?
- Allow PerplexityBot and Perplexity-User. The first builds the index, the second fetches pages while a user waits for an answer. A block on either is the block that costs a citation. Many robots files written to keep AI training crawlers out block these too, by accident.
- Serve real text in the HTML. Perplexity does not run your scripts.
- Date the page. Perplexity favours recent pages for anything that changes: prices, versions, comparisons. A visible date and a dateModified in the markup that agree with each other help it choose you over an undated page.
- Answer under the heading, in a short passage. Citations are passages. A heading that asks the question with a direct two-sentence answer beneath it gets cited; a page that circles the point does not.
- Keep one subject per page. When the title, the heading and the opening line agree, the engine knows what it is citing.
The free site check tests the crawler permissions and the quotability of your pages, and names what fails.
Condition two: are you on the pages it cites?
For a buying question, Perplexity’s sources are the same kind of pages the other engines read, with one difference of weight: it cites community threads more than any other engine. “Is X worth it” is answered from Reddit, Quora and product forums, in the words of whoever posted there.
| Where | What it does in the answer |
|---|---|
| Comparison and alternatives pages | The source for “best” and “vs” questions |
| Review sites | Cited for ratings and the phrase “users report” |
| Reddit, Quora and forums | Cited for honest-opinion questions, heavily |
| Recent articles and press | Preferred for anything with a date on it |
| Your own pages | Cited for facts about you: price, features, where you operate |
The forum threads are the ones a brand can join honestly. Answer as yourself, with your name on it, and say what your product does and does not do. Never post as a customer: the thread is read by buyers and by the engine, and both notice.
Being named and being cited are different numbers
Perplexity can cite your page and recommend somebody else, or recommend you from a review site without citing you at all. Both happen in stored sessions. The first means the engine read you and was not persuaded; the fix is on the page it cited. The second means your reputation is carried by other sites; the fix is to make sure those sites stay right about you.
Oftheard records both for every answer: whether you were named, who was recommended instead, and every source the engine cited, rolled up on the Sources page into the sites that helped a competitor and never helped you.
What does not work
- Blocking PerplexityBot to stop training, then expecting citations. It is the crawler that finds you.
- Undated evergreen pages for questions that change. Perplexity picks the page with a date.
- Posting as a customer in a thread. Moderators remove it and the engine reads the removal thread.
Where to start this week
- Check that PerplexityBot and Perplexity-User are allowed, and that your pages carry a real date.
- Ask Perplexity your buyers’ ten questions and write down every source it cites. That list is your outreach plan.
- Join the threads on it, as yourself.
- Ask the same ten questions again in a fortnight and compare the sources, rather than the wording.
Questions
Which crawler does Perplexity use?
Two. PerplexityBot builds its index, and Perplexity-User fetches a page while somebody waits for an answer. Both must be allowed in robots.txt for a page to be cited. A rule written for GPTBot does not cover them.
Why does Perplexity cite Reddit instead of my site?
Because the buyer asked for an opinion and Reddit is where opinions are written. Your site is cited for facts about you. The honest route into those threads is to answer in them under your own name.
Does Perplexity work for buyers in India?
Yes. It is asked in English from India like any other market, and its sources for an Indian buying question skew to Indian comparison pages, marketplaces and forums. Oftheard asks it through its API with search on, as a buyer in India, and stores every citation.
Can I see which of my pages Perplexity cites?
Yes, on the Sources page of a session: every page each engine cited, ranked by how often, with the ones that helped a competitor and never helped you marked.
More guides
- How to rank on Gemini and be named in its answers
- How to appear in Google AI Overviews in India
- How to rank in Google AI Mode, and be cited there
- How to get ChatGPT to recommend your business
- Free AI visibility checker: a score in half a minute
- Answer engine optimization services in India
- How to rank on ChatGPT, and what ranking even means there