--- title: "How does Perplexity choose its sources?" slug: perplexity-source-selection canonical: https://jsonburns.com/answers/perplexity-source-selection datePublished: 2026-07-23T23:39:39.844Z dateModified: 2026-07-23T23:39:39.844Z license: https://jsonburns.com/license publisher: JSON Burns cluster: B ---# How does Perplexity choose its sources? Perplexity runs live web retrieval for nearly every query, ranks the results for relevance, authority, and freshness, then writes an answer with numbered citations to the handful of sources it used. To become one, be retrievable (server-rendered, crawlable, allow PerplexityBot), answer the specific question cleanly near the top, and carry real authority signals — Perplexity cites sources, so being citable is the entire game. _Last updated Jul 23, 2026_ _Published by JSON Burns · Licensed under Citation License 1.0 (https://jsonburns.com/license)_ ## Why Perplexity is the engine to study Perplexity is the purest answer engine of the major players: it's citation-first by design. Where ChatGPT sometimes answers from memory, Perplexity runs live retrieval for nearly every query and shows its sources as numbered citations. That makes it the clearest place to learn how source selection works — and a high-value target, since it crossed tens of millions of monthly users and handles hundreds of millions of queries a month (Source: DemandSage, 2026). ## How the pipeline works Perplexity's pipeline is four steps: live retrieval for nearly every query, ranking on relevance and authority and freshness, synthesis with numbered citations, and follow-up questions that trigger fresh retrieval. *In the field —* Perplexity is where I send clients to actually SEE this working, because it shows its sources right there. We read the citations together and it turns an abstract problem into a checklist: here are the ten questions your competitors get named on and you don't. 1. **Live retrieval, almost always.** Perplexity searches the web for nearly every query rather than relying on training data, using its own index plus partner search results. 2. **Ranking.** It ranks candidates on relevance to the specific question, source authority and trust, freshness, and how cleanly the content answers the query. 3. **Synthesis with citations.** It writes a concise answer and attaches numbered citations to the few sources it actually drew from — typically a handful, not dozens. 4. **Follow-ups.** It suggests related questions, each triggering a new retrieval — so a single topic can surface your content across several follow-up answers. > **Expert Take — Jason Burns:** Perplexity is where I send clients to *see* AEO working, because it shows its work. Ask it about your category and the citations are right there — you can read exactly which competitors it trusts and which questions you're absent from. It turns an abstract problem into a checklist: here are the ten questions I should be cited on and currently am not. ## What gets you selected You get selected by being retrievable, directly answering the specific question, carrying authority signals, staying fresh, and using clean structure. Retrievability comes first. *In the field —* the most common own-goal I find on Perplexity is a site quietly blocking PerplexityBot in its robots.txt — usually a stray line copied from some old template years ago — and then the owner wonders why they're never cited. Before anyone touches a word of copy, I check that one line. I've taken sites from completely invisible to cited with a one-character change, because the content was fine all along; the crawler was just locked out at the door. Perplexity is also the engine I send clients to first, because it shows its sources right on the page. We read the citations together and it turns an abstract problem into a concrete checklist: here are the exact questions your competitors get named on and you don't. **Retrievability first.** Allow `PerplexityBot` (its index crawler) and `Perplexity-User` (its on-demand fetch) in robots.txt, serve server-rendered HTML, and keep a clean crawl path. If it can't fetch and read you, nothing else matters. **Directly answer the specific question.** Perplexity rewards pages that cleanly resolve the exact query near the top. A self-contained answer block in plain language is what gets pulled into the synthesis. **Authority and trust signals.** It weighs domain authority, corroboration across sources, and genuine expertise. Named authors with verifiable identities, real reviews, and third-party mentions raise your selection odds. **Freshness.** Current, dated content beats stale pages, especially for anything time-sensitive. **Structure.** Clean headings, lists, and schema make extraction easy. FAQPage markup packages question-answer pairs the engine can lift. > **Expert Take — Jason Burns:** The single most actionable thing with Perplexity is to allow its crawlers and confirm it can actually reach you. I've seen sites blocking PerplexityBot in robots.txt by accident — usually copied from an old template — and wondering why they're never cited. Check that one line before anything else. ## How to measure Run your real customer questions in Perplexity and log which sources it cites and whether you're among them. Because it shows citations explicitly, this is the easiest engine to audit by hand. Track inclusion (how often you appear), cited share (how often you're a numbered source), and trajectory over time. Filter your logs for `PerplexityBot` and `Perplexity-User` to confirm it's fetching your pages. ## FAQ ### How does Perplexity choose its sources? Perplexity runs live web retrieval for nearly every query, ranks the results for relevance, authority, and freshness, then writes an answer with numbered citations to the handful of sources it actually used. Being retrievable, directly answering the question, and carrying authority signals are what get you selected. ### Does Perplexity search the web for every answer? For nearly every query, yes. Unlike assistants that often answer from training data, Perplexity is citation-first and retrieves live results for most questions, which is why its sources are visible as numbered citations. ### What crawler does Perplexity use? Perplexity uses PerplexityBot to build its index and Perplexity-User for on-demand fetches. Allow both in your robots.txt — accidentally blocking PerplexityBot is a common reason sites are never cited. ### How do I get cited by Perplexity? Be retrievable (server-rendered, crawlable, allow PerplexityBot), answer the specific question cleanly near the top of the page, carry real authority signals, keep content fresh, and use clean structure and schema so the answer is easy to extract. ### How many sources does Perplexity cite? Typically a handful per answer — the few sources it actually drew from — shown as numbered citations, with related follow-up questions that each trigger fresh retrieval. --- © Adolicious LLC · By Jason Burns · Canonical: https://jsonburns.com/answers/perplexity-source-selection