How Perplexity Chooses Its Sources (and How to Become One)
Perplexity runs a live web search for every question, reads the most relevant pages it finds and writes its answer with numbered citations to the ones it used. To become one of its sources, let PerplexityBot crawl your site, publish a page that answers the exact question near the top with specific, dated facts, and get mentioned on the sites Perplexity already cites in your category, which for business software include Reddit, YouTube, LinkedIn and G2.
Key takeaways
- Perplexity searches the web for every answer, so a new page can become a source once it's crawled, without waiting for a new model.
- PerplexityBot builds Perplexity's search results and isn't used to train foundation models. Blocking it keeps you out.
- Pages that answer the question in their first lines, with specifics and a visible date, are the easiest to cite.
- In a 30-million-source study, G2 appeared among Perplexity's five most-cited sites and not among any other engine's.
How does Perplexity build an answer?
Perplexity describes itself as searching the internet in real time for each question, then summarizing what it found with numbered citations linking to the original sources. Every answer starts with a search, not with what a model remembers. That's good news if you're new: you don't have to be in a model's training data to be cited, you have to be in the results Perplexity retrieves and among the pages it chooses to use.
What decides which pages it cites?
Four things, in order:
- Can Perplexity crawl you? Perplexity's crawler for search is PerplexityBot. Perplexity says it is "designed to surface and link websites in search results on Perplexity" and "not used to crawl content for AI foundation models", and recommends allowing it in robots.txt. A separate fetcher, Perplexity-User, opens a page when a person asks Perplexity to, and Perplexity says it "generally ignores robots.txt rules" because a user requested it.
- Do you match the question? Buyers ask Perplexity full questions: "best CRM for a five-person agency that lives in Gmail". A page built around that situation is retrieved for it. A generic features page often isn't.
- Can it lift a clear answer? Perplexity cites the passages that support each sentence it writes. A page that names its answer in the first two sentences, then backs it with a table and specifics, gives it something quotable.
- Do other sources agree? Perplexity reads several pages per answer. When review sites, forums and roundups say the same thing about you, you're a safe source to cite and a safe brand to name.
Perplexity's developer API lets searches be narrowed by domain, language and recency. That's a reminder that dates and domains are attributes its search keeps track of, so show an "updated" date on pages that change.
Which sites does Perplexity cite most?
In an analysis of 30 million AI citations published by Peec AI in October 2026, Perplexity's five most-cited sites were Reddit, YouTube, LinkedIn, Wikipedia and G2. G2 appeared in Perplexity's top five and in no other engine's, which points to Perplexity leaning on review sites for business software questions.
| Source | Why Perplexity uses it | What you can do |
|---|---|---|
| First-hand experience from buyers and users | Join relevant threads as yourself, disclose who you are, answer the actual question | |
| YouTube | Reviews, walkthroughs and comparisons | Publish short, specific videos for buyer questions, with clear titles and descriptions |
| Articles and posts from practitioners | Publish useful posts and articles under real names on your team | |
| Wikipedia | Definitions and background | Only if you're genuinely notable; don't write your own article |
| G2 | Verified reviews and category grids | Keep your profile complete and current, and ask real customers for reviews |
How do you become a Perplexity source, step by step?
- Allow PerplexityBot in robots.txt. The free AI Crawler Checker shows whether your site blocks it.
- List the buyer questions you want to be named in, in your buyers' words.
- Ask each one in Perplexity and write down the brands named and the pages cited. Those pages are your map.
- Publish one page per question on your own domain: the answer first, a fair comparison table, current prices and limits, sources, an FAQ and an updated date.
- Get onto the cited pages. If Perplexity keeps citing one roundup or one G2 category, being included there is worth more than another post on your own blog.
- Ask again every week. Perplexity re-searches for every answer, so citations move as pages are published and updated.
How do you check whether it's working?
Ask your questions in Perplexity with a logged-out browser, and record the sources each answer cites over several weeks. A single answer can differ from the next, so count how often your domain and your brand appear, not whether they appeared once. Citepoint checks Perplexity weekly through its API, which searches the web for every answer like the app does, alongside ChatGPT, Gemini, Claude and Google's AI answers. The free scan shows who AI recommends for three of your buyer questions.
Frequently asked questions
Does Perplexity use my content to train its models?
Perplexity says PerplexityBot is used to surface and link websites in its search results and is not used to crawl content for AI foundation models.
If I block Perplexity-User, will Perplexity stop citing me?
Blocking PerplexityBot is what keeps you out of Perplexity's search results. Perplexity-User only fetches a page when a person asks Perplexity to, and Perplexity says it generally ignores robots.txt for those requests.
How quickly can a new page get cited?
Once PerplexityBot has crawled it and it's relevant to the question, it can be cited, because Perplexity searches for every answer. In practice that's usually days to weeks, depending on how quickly the page is discovered.
Does Perplexity only cite big websites?
No. It cites the pages that best support what it writes. A specific page on a small site that answers the exact question can be cited alongside large ones, especially when other sources agree with it.