How to Get Cited by Perplexity: Its Crawlers, Its Sources and the Pages It Quotes

An answer engine that footnotes everything

Perplexity describes itself as searching the web in real time and backing every answer with citations and links to the original sources. It offers regular Search, Pro Search and Deep Research for questions of different weight. Unlike a chatbot that answers from memory first, Perplexity reads before it writes, and it shows the reading list.

That reading list is the whole game. A Perplexity answer about "best commercial roofers in San Antonio" might cite a trade directory, a Yelp page, a local news piece and two contractors' websites. Every one of those is a door. If your site is not cited, one of the other sources may still mention you, and that counts. If none of them mention you, Perplexity has nothing to work with, however good your service is.

Perplexity also runs its own index rather than borrowing a rival's. When it opened a Search API to developers in September 2025, press reports described it as the same index of hundreds of billions of pages that powers the answer engine, priced at $5 per 1,000 requests. So the first question is simple: has Perplexity's crawler been able to read you at all? That is the subject of the next section, and of our broader generative engine optimization guide.

PerplexityBot and Perplexity-User play by different rules

Perplexity documents two user agents, and they behave differently enough that a single robots.txt rule cannot govern both. Read this table before you copy anyone's "block the AI bots" snippet.

Two consequences follow. First, blocking PerplexityBot buys you nothing on the training front, since Perplexity says that crawler is not used to build foundation models; it only costs you citations. Second, because PerplexityBot did not run JavaScript in Vercel's testing, a page whose text appears only after scripts load may look blank to it. View the page source of your key service page. If the words are missing from the raw HTML, read our guide to JavaScript and AI crawlers before writing another word of content.

Yelp, Tripadvisor and Wikipedia: Perplexity's borrowed eyes

For local businesses, the most important fact about Perplexity is not on Perplexity's site. Since March 2024, according to press reports, Perplexity has licensed Yelp data, including maps, reviews and photos, for local business answers. Since January 2025 its hotel results include Tripadvisor ratings, perks, images and summaries, again per press reports. And in January 2026 Perplexity joined Wikimedia Enterprise, the paid API for Wikipedia and Wikidata content.

Put plainly: a restaurant, dentist or contractor that has ignored Yelp because "nobody uses Yelp anymore" has ignored the listing Perplexity is reported to read. A Peec AI analysis of 30 million AI citations in March 2026 also ranked Yelp among the ten most-cited domains across five AI engines, with Reddit first; that study comes from a tool vendor, so treat it as a signal rather than a census. Our page on Reddit, forums and directories covers the rest of that list.

• Claim and complete the Yelp page. Categories, hours, service area, photos and a description that uses the same name, address and phone as your site and Google profile.
• Do not ask for Yelp reviews. Yelp's own business site says asking for reviews or offering incentives is against its policy, and its recommendation software is built to spot solicited reviews and not recommend them. Yelp suggests check-ins, "Find us on Yelp" signage and good service instead.
• Reply to the reviews you have. A calm, specific owner reply is public text that sits next to the complaint wherever the review is read.
• Leave Wikipedia to independent editors. Wikipedia strongly discourages editing with a conflict of interest and requires paid editing to be disclosed. If you have an article, suggest corrections on its talk page.
• Hotels: tend Tripadvisor the way restaurants tend Yelp, since press reports say it feeds Perplexity's hotel results.

Reviews affect every engine differently, and the rules differ by platform. Our guide to reviews and AI recommendations lines them up, including where Google's policy is stricter than the FTC's.

Writing the paragraph Perplexity lifts into its answer

Perplexity is the live engine the original GEO research was tested on. The 2023 paper from Princeton University and IIT Delhi ran most of its experiments on a simulated engine, but on live Perplexity it found that adding quotations improved one visibility metric by 22% over baseline. Across the study, the strongest methods were adding quotations, adding statistics, citing sources and improving fluency. Keyword stuffing did little or nothing. The authors also note their methods may need to change as engines evolve, so read it as direction, not a recipe.

Perplexity cites pages that sound like evidence. Write the sentence a reporter would want to quote, then put your name next to it.

Here is what that looks like on a working business site. None of it is exotic, and all of it serves human readers as well as machines.

• Answer first. Open each service page with two or three sentences that answer the obvious question outright: what it costs, how long it takes, who it is for.
• Use real numbers you can stand behind. "Most bathroom remodels we finish run 3 to 5 weeks" beats "fast turnaround", provided it is true for your jobs.
• Name your sources. Cite the code section, the manufacturer's warranty terms or the regulator's page, with a link. That is the "cite sources" method in practice.
• Quote someone accountable. A short, attributed quote from your licensed technician or your lead estimator, about a specific job condition, gives the engine a quotable line.
• Keep facts current. A price page from 2022 is a liability. Update it when prices change, and say when it was last reviewed. More in content freshness.
• Use tables for comparisons. Options, prices and specs in a table are easier to lift intact than the same facts buried in a paragraph.

The deeper version of this, with before-and-after examples, is our guide to content AI can quote.

How to check what Perplexity says about you

Perplexity makes self-auditing easier than any other engine because the sources are always on screen. The job is to turn those sources into a list of targets.

1. 1

   Collect the questions

   Write 10 to 20 questions customers actually ask, with a place in each: "which HVAC companies in Schertz do same-day repairs", "is it worth fixing a 15-year-old water heater". Our prompt tracking guide covers sample size.

2. 2

   Run each one in Perplexity

   Use the same mode every time, since Search, Pro Search and Deep Research are built for different depths of research. Record whether your business is named and in what words.

3. 3

   Log every cited source

   Write down each domain in the source list. Mark which ones mention you, which mention competitors only, and which mention nobody.

4. 4

   Sort the gaps

   A directory that lists competitors but not you is a listing to claim. An article that names three rivals is a pitch to make. A competitor's page cited for a question you answer better is a page to write.

5. 5

   Check the facts it states

   Compare hours, prices, service area and phone against reality. Note the source behind any error; that is where the fix goes.

6. 6

   Repeat monthly

   Keep a dated log. Our free AI visibility check runs questions through several engines through their APIs, which can answer slightly differently from the apps, so use both.

Perplexity beside ChatGPT and Google AI Overviews

Each engine reads a different slice of the web. Work done for one carries over partly, not fully. The rows below are the ones that change what you do on Monday morning.

The full engine pages go deeper: ChatGPT, Google AI Overviews, Gemini and Microsoft Copilot.

Correcting Perplexity, counting its visits, selling through it

When the answer is wrong

Because Perplexity cites its sources, a wrong answer usually has a visible culprit: an outdated directory listing, an old press release, a competitor's comparison page. Fix that source or ask its owner to. Then report the answer: Perplexity's help center says to open the three-dot menu below the response and select Report. Retest after the source changes. Our guide to AI citing old information covers stubborn cases.

When you want to see the traffic

Perplexity visits can arrive from both perplexity.ai and www.perplexity.ai, so match both. Industry write-ups say Perplexity is not confirmed as part of GA4's built-in AI Assistant channel, though GA4's Source Group dimension, added in June 2026, groups Perplexity's source values together. A custom channel with a referrer rule placed above Referral catches it either way; the regex is in our GA4 AI traffic guide. App visits that send no referrer will still land in Direct.

When you sell products

On November 25, 2025 PayPal and Perplexity launched Instant Buy for US users. PayPal's Store Sync makes merchant catalogs discoverable in Perplexity, and the retailer stays merchant of record. If you already take PayPal, ask about Store Sync before buying anyone's "AI shopping" add-on. Our ecommerce GEO guide covers the other engines' checkout programs.

Perplexity work inside a DataCram program

Perplexity is one of the engines in every DataCram GEO program. We run your customers' questions through it monthly and log the sources it cites; make sure PerplexityBot can read server-rendered pages; clean up Yelp and the directories it leans on so your name, address and phone match everywhere; write answer-first pages with numbers you can defend; and pursue mentions in the articles and lists it already trusts. Our AICiteBeam platform records which sources each engine cites and alerts us when your visibility moves.

When you should skip us: if Perplexity already cites your site for your main questions and your Yelp page is in order, a monthly self-test may be all you need. Read why there is no number one in an AI answer before anyone sells you a ranking.