What is Perplexity SEO?
Perplexity SEO is the practice of structuring web content so that Perplexity AI cites your page as one of the 5-10 attributed sources it returns per query. Perplexity uses its own crawler (PerplexityBot) and its own search index — not Bing, not Google — so optimization signals translate from classic SEO partially but not fully.
Two structural facts shape the optimization game. First, Perplexity cites 5-10 sources per answer (versus ChatGPT's 3-5) — roughly double the citation slots per query, which makes Perplexity the highest-base-rate AI surface to optimize for. Second, Perplexity's authority graph is far flatter than Google's: a well-structured page on a mid-authority site can out-cite a top-3 Google ranking from a 20-year-old domain. The optimization lever is per-page content quality, not per-domain compounded authority.
Perplexity was founded in 2022 by Aravind Srinivas and surpassed 230 million queries per month in Q3 2024 (Srinivas, October 2024). The default search runs on Sonar (Perplexity's in-house Llama-3.1-based model) and Pro Search ($20/mo) lets users pick GPT-4o, Claude 3.5 Sonnet, or other frontier models per session. The optimization signals differ slightly between modes — covered in section 04.
Which Perplexity bot does what?
Perplexity ships two crawlers. Both must be explicitly allowed for the engine to consider your page; blocking either one is a silent reduction in citation share that operators rarely catch.
| Bot | Purpose | Action |
|---|---|---|
| PerplexityBot | Crawls the public web continuously to build Perplexity's own search index. This is the bot that determines whether you appear in the candidate set for any Perplexity query. | ALLOW explicitly with "User-agent: PerplexityBot" + "Allow: /". Blocking it removes you from every Perplexity query. |
| Perplexity-User | Fires when a Perplexity user asks the engine to read a specific URL in real time — on-demand, per-query fetch. | ALLOW. Blocking means a user trying to cite YOUR page through Perplexity gets a failed fetch and the engine silently picks someone else. |
How do Sonar (default) and Pro Search differ?
Free Perplexity uses Sonar by default — Perplexity's in-house Llama-3.1-based model family. Pro Search ($20/mo, also via API) lets the user pick the model per query. Each surface has slightly different source preferences; a page that wins on both has covered the full structural surface area.
Default Search (free tier)
ModelSonar — Perplexity's in-house model family, based on Llama 3.1, fine-tuned for citation-grounded answers.
Signal deltaHeavier weighting on recency and structured citations within source. Sonar is more aggressive about discarding sources that lack explicit dates or attributions.
Pro Search ($20/mo or via API)
ModelUser-selectable: GPT-4o, Claude 3.5 Sonnet, Grok, Sonar-Pro. The user chooses the model per query inside their Pro session.
Signal deltaEach Pro-backed model has its own citation preferences. Claude 3.5 favors longer-context authoritative articles; GPT-4o favors structured FAQ-style content. A page that wins on both has dense paragraphs AND clear FAQ schema.
Which Perplexity SEO signals matter most?
Ordered by leverage. Signals 1 and 2 are the floor — blocking either bot zeros your citation share. Signals 3-8 compound on top in roughly the order shown.
1. PerplexityBot allowed in robots.txt
PerplexityBot is the crawler that builds Perplexity's own search index. Unlike ChatGPT (which falls back to Bing), Perplexity has no fallback backend — if PerplexityBot can't reach you, you do not exist for any Perplexity query. Add an explicit allow rule; do not rely on the wildcard "User-agent: *" default.
2. Perplexity-User allowed for on-demand reads
When a Perplexity user pastes a URL or asks the engine to verify a specific source, Perplexity-User fetches the page live. Blocking it is the most embarrassing failure mode — the user actively chose your URL and you sent a 403. Allow it side by side with PerplexityBot.
3. Recency — visible "Last updated" + matching dateModified
Perplexity weights freshness more aggressively than any other major engine, including ChatGPT. Sources without visible dates get marginalized for time-sensitive queries — and Perplexity treats nearly all commercial-intent queries as time-sensitive. A visible "Last updated 2026-05" stamp matched with schema.org dateModified is non-negotiable on every authoritative page. Re-publish + bump the date on your top 10 pages every 60 days.
4. Citation density inside your own content
Perplexity's source-quality ranker is built around a recursion: a source that itself cites authoritative sources is more trustworthy than a source that asserts without backing. Every claim of fact should link to a DOI, .gov, .edu, or major-news source. Three attributed external citations per important page measurably shifts your Perplexity citation share — and the same investment lifts ChatGPT and Claude citation too.
5. Clean H2/H3 hierarchy — Perplexity extracts by section
Perplexity's chunker boundary-detects on heading hierarchy and prefers to cite an entire section over a sub-paragraph. Pages with shallow H1-then-paragraphs structure get extracted poorly; pages with clear H2 → H3 → paragraph hierarchy give the chunker clean cut points and produce higher-confidence cites. Treat heading hierarchy as content structure, not decoration.
6. Bullet and numbered lists — extractable formats Perplexity loves
Perplexity's answer composition heavily favors content presented as numbered lists, bulleted lists, or comparison tables. The model can lift the list verbatim and present it as part of the answer with the source attributed. Prose-only pages get fewer cites for list-style queries ("top 5 X", "steps to Y", "differences between A and B") even when they answer the question accurately.
7. Multiple distinct claims per page
Because Perplexity cites 5-10 sources per answer, a single page can be cited multiple times for distinct facts within the same answer. Pages with 8-15 distinct verifiable claims (statistics, dates, attributed quotes, comparison data points) get cited more often than equally-long pages with one central thesis. Density of distinct citable facts per 1,000 words is a measurable signal.
8. TTFB under 2 seconds — Perplexity crawler is impatient
PerplexityBot has a tighter response-time budget than Bingbot or Googlebot. Pages that take longer than 2 seconds to first byte get sampled less frequently and dropped from the index faster when the crawler is rate-budgeting. Optimize TTFB at the origin — caching, CDN edge serving, no synchronous database calls in the request path.
How do I know if I'm getting cited by Perplexity?
- Referrer = perplexity.ai in your analytics — the canonical outbound referrer
- Manual probe: 20-50 category queries on perplexity.ai, count citations
- Our free AI Citation Checker — probes Perplexity + ChatGPT + Claude for your keywords
- Per-page citation share — sample 10-20 queries per important page
- Recency-decay tracking — measure citation share weekly after a date-bump
- Direct-answer / claim-density scoring in our free audit predicts citation lift
Perplexity SEO frequently asked
How is Perplexity SEO different from Google SEO?
Two structural differences. (1) Perplexity cites 5-10 sources per answer where Google's SERP shows ten organic results — so the per-citation slot is more valuable but the bar to occupy one is lower per query. (2) Perplexity's authority graph is far flatter than Google's: a well-structured page on a mid-authority site can out-cite a top-3 Google ranking from a 20-year-old domain. The lever for Perplexity is per-page content quality; for Google it is per-domain authority compounded over years.
Does Perplexity use Bing or Google as its search backend?
Neither, primarily. Perplexity built and maintains its own crawler (PerplexityBot) and its own search index. Historically it used third-party search APIs as a complement; today the proprietary index is the primary candidate set for most queries. This is structurally different from ChatGPT, which leans on Bing — and it is why Bing SEO does NOT transfer to Perplexity in the way it transfers to ChatGPT.
Which bots does Perplexity use to crawl my site?
Two. (1) PerplexityBot — the background crawler that maintains the search index; visits your site on a recurring schedule. (2) Perplexity-User — the on-demand fetcher that fires when a user asks the engine to read a specific URL in their session. Both should have explicit allow rules in robots.txt. Many sites accidentally block Perplexity-User by treating it as a generic user agent in their bot-management tooling — check WAF / CDN bot rules in addition to robots.txt.
Why does Perplexity cite 5-10 sources per answer? Can I get into more than one slot?
Yes — for a single answer Perplexity can cite your domain multiple times if your page has multiple distinct verifiable claims. The engine composes the answer by stitching attributed claims; the same source can be the attribution for several distinct sentences within the answer. This is why claim density per page matters: 8-15 distinct attributable facts on a page measurably outperforms 1-2 even when total content length is identical.
Does Pro Search use different signals than the free Sonar default?
Yes — each Pro-backed model has its own source preferences. Claude 3.5 Sonnet favors longer-context authoritative articles with paragraph depth; GPT-4o favors structured FAQ-style content with clear answer extractability; Sonar (default) weights recency and citation density most heavily. Pages that perform on the full range need depth AND structure AND freshness — there is no single-format winner across Pro Search modes.
How do I know if Perplexity actually cites my site?
Three measurements. (1) Referrer = perplexity.ai in your analytics — the canonical referrer for outbound clicks from Perplexity cited links. (2) Manual probe — run 20-50 category queries against perplexity.ai and count citation appearances. (3) Our free AI Citation Checker — probes Perplexity (along with ChatGPT, Claude, etc.) for your brand keywords and reports citation share per engine.
Does Perplexity respect llms.txt?
Yes — Perplexity is one of the most explicit adopters of llms.txt as of 2026. Their crawler reads /llms.txt as a primary source-prioritization signal when deciding which of your pages to weight in citation candidacy. Adding llms.txt is the single highest-ROI Perplexity-specific optimization most sites have not made yet.
Why does my page rank top-3 on Google but Perplexity never cites it?
The most common cause is missing structural signals that Perplexity weights independently of Google rank — no FAQPage schema, no visible dateModified, prose-only content with no extractable lists, or PerplexityBot blocked at the CDN. Google's algorithm rewards domain authority compounded over years; Perplexity's evaluates each page on its current structural properties. A page that wins on Google can lose on Perplexity if it was written for human readers without thought to AI extraction.