← All articles

How Does Perplexity Choose Its Sources? What Its Own Docs Say (and What They Don't)

By Rei · CiteMonthly · Published 14 October 2026

Short answer: Perplexity does not publish a ranking formula for the sources it cites. What it does document is the plumbing: a crawler called PerplexityBot that is "designed to surface and link websites in search results on Perplexity" [1]; a search system that returns "real-time, ranked web results" [3]; and answers in which numbered citations like [1] map back to specific search results [2]. Its developer docs also show the levers that shape which results come back: freshness (publication and last-updated dates), domain allow/deny lists, and the searcher's location [4] [5] [6]. For a small business, that points to three things you can influence: be crawlable, be clearly dated and current, and be specific about where you operate. This article sticks to what we could open and check on 14 Oct 2026.

Where do Perplexity's sources come from?

From the live web, fetched in two ways. Perplexity's crawler page lists two user agents [1]:

  • PerplexityBot surfaces and links websites in Perplexity search results. Perplexity says it "is not used to crawl content for AI foundation models." To appear in results, Perplexity recommends "allowing PerplexityBot in your site's robots.txt file and permitting requests from our published IP ranges."
  • Perplexity-User handles user actions: when someone asks a question, Perplexity "might visit a web page to help provide an accurate answer and include a link to the page in its response." Because a person triggered the visit, "this fetcher generally ignores robots.txt rules."

Changes to your robots.txt settings may take "up to 24 hours" to be reflected [1].

So the first filter is simple: if PerplexityBot cannot reach your pages, Perplexity has nothing of yours in its index to rank. We covered how to check robots.txt and firewall settings in Is your website blocking ChatGPT?. The same checks apply here.

How does Perplexity turn search results into citations?

Perplexity's developer documentation describes how its answer engine connects text to sources. Its Agent API (the developer version of answer generation) runs web searches, receives search results, then writes text with numbered markers. In Perplexity's words, "Citation references appear in the text as numbered markers like [1], [2], mapping to the search result id field" [2]. Each search result carries an id, title, url, snippet and date [2].

Two details matter for site owners:

  1. A citation is a search result first. The docs tell developers: "Never ask the model to generate source URLs. Always use the search_results output... The search results contain verified URLs from real web searches" [2]. Put plainly, to be cited, your page has to show up in the search step.
  2. Bigger questions use more searches. "Single-step presets like fast typically search once, but multi-step presets (such as deep research) run many searches" [2]. A broad question gives more pages a chance to be retrieved. A quick one may rely on a single search.

The web search tool also limits how much it reads. Developers can set max_results per search call and cap the content "extracted from each result page" with max_tokens_per_page [4]. Defaults apply when these are omitted, and Perplexity does not publish the default values on that page.

Our inference (not Perplexity's wording): if only part of each page is read, the answer to the question should appear early and plainly on the page. A service page that buries its price range or area under three screens of introduction gives the extraction step less to work with.

Does freshness affect which sources Perplexity uses?

The docs treat freshness as a first-class control. Search results include a date and a last_updated field [3]. The API offers three kinds of date filter [5]:

  • Publication date (search_after_date_filter / search_before_date_filter), based on when content was "originally created or published."
  • Last updated (last_updated_after_filter / last_updated_before_filter), based on when content was "last modified or updated." Perplexity lists uses such as "Exclude stale content that hasn't been updated recently."
  • Recency (search_recency_filter) with values "hour", "day", "week", "month" or "year."

These are documented API parameters. Perplexity does not say which of them its consumer app applies to any given question, so do not read them as proof that newer pages always win. They do show that Perplexity's search system records a publication date and an update date for pages, and that "stale" content is something it lets developers filter out.

What to do: put a visible "last updated" date on pages that change (prices, opening hours, service areas), and actually update them. Keep old promotions off your live pages.

Can location change which local businesses get cited?

For local businesses this may be the most useful part of the docs. The Search API accepts a country code "to get more geographically relevant results," which Perplexity says is useful "for local news, regulations, and other region-specific queries" [3]. The Agent API's web search goes further with a user_location object: country, region, city, latitude and longitude. The docs add that "city and region significantly improve location accuracy" [4].

Again, that page is for developers. It does not tell us how Perplexity's app decides a user's location. It does confirm that its search layer can rank results with city-level context.

Our inference: if location context can change results, pages that state your city, neighbourhoods and service area in plain text have a better chance of matching. "We serve the greater area" is not enough. Write out the actual town names you cover.

Can Perplexity prefer or exclude particular websites?

Developers can. The search_domain_filter parameter limits results to chosen domains (allowlist) or excludes them with a - prefix (denylist), up to 20 entries, but "not both simultaneously" [6]. It can filter by domain, by top-level domain such as .gov or .edu, or by a path on a site [6]. The Agent API example excludes reddit.com and pinterest.com [4].

This matters indirectly. Apps built on Perplexity's API can be set to trust or skip certain sites. Your listing on a respected directory or local news site may reach an answer that your own domain does not, and the reverse is also true.

What does Perplexity not tell us?

Being clear about gaps is part of fact-checking. In the pages we opened, Perplexity does not publish:

  • a list of ranking factors or their weights;
  • whether reviews, backlinks or schema markup affect citation;
  • any way to pay for organic citation, or to ask for it;
  • the default max_results or page-extraction budgets used by its consumer app.

Perplexity's Help Center has articles on how answers are built and on "source labels." We could not open them on 14 Oct 2026 (they timed out or returned 403 to our tools), so we have not quoted them here. If you see claims online about "Perplexity's top 3 ranking factors," check whether the author links to a Perplexity page that says so.

A practical checklist for small businesses

  1. Allow PerplexityBot in robots.txt and let its published IP ranges through your firewall or CDN [1]. If you use a web application firewall, Perplexity's page shows a Cloudflare rule that matches the user agent and IP range, with action Allow [1].
  2. Wait about a day after changing robots.txt, then test [1].
  3. Answer the question early on the page. Lead with the facts customers ask about: what you do, where, for whom, and price ranges if you publish them. (Our inference from the per-page extraction limits [4].)
  4. Date your pages honestly and update them. Perplexity's results carry date and last_updated fields [3] [5].
  5. Name your locations in text. City and region are documented location signals in the search API [4].
  6. Be present on other credible sites (local press, trade bodies, directories) because answers are built from many search results, and domain filters can favour some sites over others [2] [6].
  7. Test it yourself. Ask Perplexity the questions your customers ask, note which sites it cites, and look at what those pages do that yours does not.

Frequently asked questions

Does Perplexity train its AI models on my site if I allow PerplexityBot?

According to Perplexity, PerplexityBot "is not used to crawl content for AI foundation models" [1]. It is for search results.

If I block PerplexityBot, can Perplexity still link to my site?

Possibly. Perplexity-User fetches pages when a user asks and "generally ignores robots.txt rules" [1]. Blocking PerplexityBot removes you from its crawled search results. It does not necessarily stop every visit.

Is there a Perplexity equivalent of Google Search Console?

We found no such tool in the pages we opened. The practical check is to look in your server or CDN logs for PerplexityBot requests, and to run your own test questions.

Does newer content always beat older content?

Not as far as anyone can show. The docs show that publication and update dates are recorded and can be filtered on [5], not that the app always prefers the newest page.

Want to know what Perplexity says about you?

CiteMonthly runs a free "Does ChatGPT mention you?" check. We ask ChatGPT and Perplexity the questions your customers ask, then email you their answers: who gets named, which sources they cite, and whether you're one of them. No card, no account. Email citemonthly@gmail.com to request your free check.

Sources

  1. Perplexity docs, "Perplexity Crawlers". https://docs.perplexity.ai/guides/bots (opened 14 Oct 2026)
  2. Perplexity docs, "Streaming Citation Parsing". https://docs.perplexity.ai/docs/cookbook/articles/streaming-citations/README (opened 14 Oct 2026)
  3. Perplexity docs, "Perplexity Search API" (quickstart). https://docs.perplexity.ai/docs/search/quickstart (opened 14 Oct 2026)
  4. Perplexity docs, "Web Search" (Agent API tool). https://docs.perplexity.ai/docs/agent-api/tools/web-search (opened 14 Oct 2026)
  5. Perplexity docs, "Search Date and Time Filters". https://docs.perplexity.ai/docs/search/filters/date-time-filters (opened 14 Oct 2026)
  6. Perplexity docs, "Search Domain Filter". https://docs.perplexity.ai/docs/search/filters/domain-filter (opened 14 Oct 2026)