Most AI assistants answer questions without sending you any traffic. Perplexity is the exception: every answer it gives includes numbered citations with links, and people actually click them. That makes Perplexity one of the few AI platforms where "AI SEO" translates directly into visits. To get cited, you need three things: let Perplexity's crawler reach your content, write in a format its system can extract and quote, and be the kind of source worth citing — specific, current, and trustworthy. This guide shows you how to do all three.
How Perplexity chooses what to cite
When someone asks Perplexity a question, the system searches the web, retrieves a set of relevant pages, and writes an answer grounded in those pages. The citations you see — the little numbered links — point to the sources it actually used. Your goal is to be one of those sources.
Perplexity doesn't publish a ranking formula, but observing which pages get cited reveals a consistent pattern. Cited pages tend to:
- Answer the specific question directly, near the top of the page.
- Contain concrete facts: numbers, dates, names, steps — things a summary can quote.
- Be current. For anything time-sensitive, recently updated pages win.
- Show clear authorship or organizational accountability, so the answer has someone standing behind it.
- Be technically accessible: fast, crawlable, and not blocked to the crawler.
In other words, the pages Perplexity cites are the pages a careful human researcher would bookmark. There is no trick here — but there is a craft to being quotable, and most sites aren't doing it.
Step 1: Make sure Perplexity can crawl you
Perplexity uses two documented user agents. PerplexityBot is its search crawler — it discovers and indexes content for Perplexity's search results, and Perplexity states it is not used for AI model training. Perplexity-User is a fetcher triggered when a user asks Perplexity to look at a specific page.
If you want citations and the traffic that comes with them, don't block PerplexityBot. Check your robots.txt for lines like these:
User-agent: PerplexityBot
Disallow: /That rule tells Perplexity's crawler to stay out entirely — which also means your pages can't be discovered for citations. If you find it, decide deliberately whether you want it there. Many sites block AI crawlers reflexively with broad rules and never revisit the decision. If referral traffic matters to you, PerplexityBot is one of the bots worth allowing:
User-agent: PerplexityBot
Allow: /You can build and check a proper robots.txt with a robots.txt generator, then verify the result with a robots.txt tester to make sure your rules do what you think they do. Google's own robots.txt documentation is the authoritative reference for the syntax.
One honest caveat: Perplexity-User, the user-triggered fetcher, may visit a page when a person explicitly asks about it, and Perplexity has said this fetcher generally doesn't follow robots.txt because the fetch was requested by a user. Also be aware that Cloudflare publicly accused Perplexity in 2025 of evading blocks by rotating user agents and IPs — Perplexity disputed the characterization, but the episode is a reminder that robots.txt is a voluntary standard, not a lock. The practical takeaway: allow PerplexityBot if you want the citation traffic; robots.txt alone won't fully control determined fetching.
Step 2: Write content Perplexity can extract and quote
Perplexity's system reads your page and lifts passages into its answer. Pages that are easy to lift from get cited more. Here's what that means in practice:
Lead with the direct answer
Put the core answer in the first paragraph. If the page is "How to block GPTBot," the first paragraph should say how to block GPTBot — not three paragraphs of background about AI crawlers. Perplexity quotes what it can find quickly, and so do human readers.
Use question-shaped headings
Structure your H2 and H3 headings as the questions people actually ask: "How do I create an llms.txt file?", "What is the difference between robots.txt and llms.txt?", "Does blocking Google-Extended stop AI Overviews?" (It doesn't, by the way — Google-Extended only opts content out of Gemini model training, not AI Overviews, which are built from Google's search index.) When your headings mirror real queries, Perplexity's retrieval system matches your sections to questions more reliably.
Make facts easy to lift
AI systems quote sentences, not vibes. Compare:
- Weak: "Sitemaps should generally be kept reasonably sized."
- Strong: "The sitemap protocol allows up to 50,000 URLs or 50 MB uncompressed per sitemap file."
The second version is a citable fact. Write more sentences like the second one: specific numbers, named entities, dates, and defined terms.
Use tables for comparisons
When you compare options — tools, plans, methods — put the comparison in a table. Tables are one of the most reliably extracted formats. A paragraph comparing five tools will get summarized; a table comparing five tools will get quoted.
Keep it scannable and current
Short paragraphs, bullet points for lists, numbered steps for processes. And date your content: "Updated October 2026" signals freshness to both readers and retrieval systems. An undated page with 2022 statistics will lose citations to an updated competitor even if the rest of the content is equal.
Step 3: Become the kind of source worth citing
Extraction gets you considered; authority gets you chosen. When Perplexity has ten pages saying similar things, it cites the ones that look most trustworthy:
- Cite your own sources. Link to primary sources — official documentation, original studies, named experts. Pages that show their work get cited more than pages that assert things.
- Show who wrote it. A named author with a bio beats an anonymous page. This is especially true for advice content where the author's experience is the credential.
- Publish original material. Your own test results, screenshots, measurements, and case data can't be found anywhere else. When Perplexity needs a fact only you have, you're the citation.
- Keep claims verifiable. Avoid superlatives you can't prove ("the best," "the fastest") and statistics without sources. Systems — and readers — discount pages that overclaim.
- Update on a schedule. Review your most-cited-potential pages quarterly. Fresh dates and current numbers are a quiet but powerful citation advantage.
An llms.txt file can also help AI systems discover your most important pages faster by giving them a structured summary of your site — worth adding once your content is in good shape.
Step 4: Measure your Perplexity traffic
Here's the good news: unlike most AI platforms, Perplexity sends traffic you can actually measure. When someone clicks a citation, they arrive at your site as a normal visitor with the referrer perplexity.ai.
In Google Analytics 4, go to Reports → Acquisition → Traffic acquisition and look for perplexity.ai in the session source/medium table. To track it ongoing, create a custom segment or an exploration filtered to sessions where the source contains "perplexity". Watch sessions, engagement rate, and conversions from this segment separately — it behaves differently from Google organic traffic, and it's growing for most publishers who optimize for it.
A quick note on what analytics can't show: GA4 only sees visitors who load your page with JavaScript. It cannot show you bot crawls — no analytics tool can, because crawlers don't execute JavaScript. So don't look for "PerplexityBot visits" in GA4; look for human referral traffic from perplexity.ai. For crawler activity, check your server logs instead.
Beyond analytics, do a monthly spot check: ask Perplexity five to ten questions your content answers, and note whether your site gets cited. Track it in a spreadsheet. Citation share is the KPI that matters most for this channel, and nobody will report it to you automatically.
What doesn't work
A few tactics people try that waste effort or backfire:
- Keyword stuffing for AI. Repeating phrases doesn't make a page more citable. Clear, factual writing does.
- Blocking PerplexityBot while expecting citations. If the crawler can't discover your pages, you rely entirely on user-triggered fetches — a much weaker position.
- Publishing "Perplexity bait" with no substance. Thin pages optimized for extraction get ignored in favor of pages with real depth. The extraction-friendly format only works on top of genuine content.
- Hiding content from crawlers but showing it to users. Cloaking violates search engine guidelines and can get you penalized by Google too. Keep one version of the truth.
- Expecting instant results. Crawling, indexing, and citation patterns take weeks to shift. Give changes at least a month before judging them.
Your Perplexity citation checklist
- Confirm PerplexityBot is allowed in robots.txt (unless you've deliberately decided otherwise).
- Put the direct answer to the page's main question in the first 100 words.
- Rewrite headings as real user questions.
- Replace vague claims with specific, sourced facts.
- Add tables for any comparison on the page.
- Date the page and schedule quarterly freshness reviews.
- Add a named author with a real bio.
- Link out to primary sources for key claims.
- Set up a GA4 segment for perplexity.ai referral traffic.
- Run a monthly citation spot-check with real questions.
Perplexity rewards exactly what good publishing has always rewarded: clear, specific, trustworthy content — just with the added requirement that machines can lift it cleanly. Do the fundamentals well, keep the crawler welcome, and measure what arrives. For a broader view of how AI search differs from classic SEO, see our guide to SEO vs. AI search. OpenAI documents its own crawler controls at platform.openai.com/docs/bots if you want to compare how different AI companies handle bot access.
Frequently asked questions
Does Perplexity really send traffic, unlike other AI assistants?
Yes. Perplexity's answers include visible numbered citations that link directly to source pages, and users click them. It's one of the few AI search products where publishers consistently report measurable referral traffic. Most other assistants either don't link sources or bury them.
Should I allow or block PerplexityBot?
If you want citation traffic, allow it. Perplexity states PerplexityBot is a search crawler not used for model training. Blocking it removes your pages from Perplexity's index. The only strong reason to block is a principled objection to AI crawling in general — in which case you'd likely block other AI crawlers too.
What's the difference between PerplexityBot and Perplexity-User?
PerplexityBot is the automatic search crawler that builds Perplexity's index. Perplexity-User fetches a specific page when a user asks Perplexity about it. The user-triggered fetcher generally doesn't follow robots.txt, since a person explicitly requested the page.
How long does it take to start getting cited?
After making your content citable and crawlable, allow four to eight weeks. Perplexity needs to crawl and index the updated pages, and citation patterns shift gradually as its system encounters your content for real queries.
Can I see which of my pages Perplexity cites?
There's no official dashboard. The practical method is manual: regularly ask Perplexity questions in your niche and record citations. Combine this with your perplexity.ai referral segment in GA4 to connect citations to actual visits.
Does llms.txt help with Perplexity citations?
llms.txt helps AI systems discover and understand your site's structure, which supports the crawlability side. But citations are earned by the content of individual pages — clear, factual, well-structured writing matters far more than any single file.
