Does Perplexity train AI models on my site?

Perplexity says PerplexityBot is for search and isn't used to crawl content for AI foundation models, and that it doesn't build foundation models itself.

Can I stop Perplexity from reading my pages?

You can disallow PerplexityBot in robots.txt to stay out of its index. Perplexity-User, which fetches pages when a user asks, generally ignores robots.txt according to Perplexity, so blocking it needs a firewall rule. Blocking either one makes it much less likely you're cited.

Why does Perplexity cite a directory instead of my website?

Usually because the directory page answers the question more directly (a list of options with ratings) or your own page doesn't mention the service and location in plain text. Being listed on those directories, and having a specific page of your own, both help.

How fast does Perplexity pick up changes?

Robots.txt changes apply within about 24 hours, according to Perplexity. New or updated pages are picked up when they're recrawled; links from other sites help them get found sooner.

How to show up in Perplexity: a small-business guide

Perplexity runs a web search for every question and shows its sources. If your pages answer the question clearly, you can be one of them.

Updated 21 Oct 2026 · 2 official sources

How does Perplexity find businesses?

Perplexity is an answer engine: for each question it searches the web in real time, picks sources, and writes an answer with numbered citations. Its index is built by its own crawler, PerplexityBot, and Perplexity says it also partners with third-party crawlers to build that index.

That makes Perplexity the most "search-like" of the assistants. Pages that answer one question clearly, are easy to fetch, and are backed by other sites tend to get cited. Because the citations are visible, you can see exactly which sources beat you and why.

Which crawlers matter for Perplexity?

Perplexity crawlers and what to do with them
User agent	What it does	Do
PerplexityBot Search Indexes pages so Perplexity can surface and link them. Perplexity says it is not used to train AI foundation models.	Indexes pages so Perplexity can surface and link them. Perplexity says it is not used to train AI foundation models.	Allow
Perplexity-User User-triggered Fetches a page when a user's question needs it. Perplexity's docs say this fetcher generally ignores robots.txt, since a person asked.	Fetches a page when a user's question needs it. Perplexity's docs say this fetcher generally ignores robots.txt, since a person asked.	Allow

robots.txt you can copy

User-agent: PerplexityBot
Allow: /

User-agent: Perplexity-User
Allow: /

Not sure what your file says now? Check it free: it reads your robots.txt rule by rule for PerplexityBot, Perplexity-User and 19 other AI bots.

What should you fix, step by step?

01Allow PerplexityBot, in robots.txt and in your firewall
Make sure robots.txt doesn't block PerplexityBot (check with the AI Crawler Checker). If you use a WAF or bot protection, Perplexity publishes its crawler IP ranges so you can allowlist them. Perplexity says robots.txt changes can take up to 24 hours to apply.
If a page is blocked, Perplexity says it may still index the domain, the headline and a brief factual summary, but not the page text, so you lose the detail that gets you cited.
02Answer specific questions on specific pages
Perplexity cites pages, not businesses. A page titled "Emergency boiler repair in Bristol: prices, response times and what to expect" is far easier to cite than a homepage that says "quality service since 1998". Write one page per service and answer the real questions: cost ranges, how long it takes, areas covered, what's included.
Use clear headings, short paragraphs and lists. Put a visible "last updated" date on pages with prices or hours, and keep it true.
03Put your key facts in plain HTML text
PerplexityBot reads the HTML your server sends. Facts that only exist inside images, PDFs, sliders or text loaded later by JavaScript are easy to miss. Your homepage and service pages should say, in ordinary text: what you do, for whom, where (city and areas served), how to book, hours, and typical prices if you can share them.
One page per main service, with the questions customers actually ask answered on it, gives an assistant something specific to quote. Run the SEO Health Snapshot for the basics, and see AI crawlers and JavaScript.
04Add LocalBusiness schema that matches your listings
Structured data (JSON-LD) states your facts in a form machines don't have to interpret: business type, address, phone, opening hours, area served, price range and your profile links in sameAs. It doesn't guarantee a mention, but it removes ambiguity, and it must say exactly what your listings say.
Use the most specific type that fits (Dentist, Plumber, Attorney…). Our free Schema / JSON-LD Generator writes it for you.
05Earn real reviews where this engine looks
When someone asks for "the best" or "a trusted" business, assistants lean on review platforms and third-party lists rather than your own claims. Look at the citations Perplexity shows for "best [your service] in [your city]". Review platforms, local directories and roundup articles that appear there are the third-party sources worth being on.
Ask happy customers for reviews as part of your normal process, reply to reviews (good and bad) and fix the issues they raise. Never buy or script reviews: platforms remove them, and it can get a listing suspended.
06Make your name, address and phone identical everywhere
AI answers are stitched together from many sources at once: your site, your listings, review sites and directories. When they disagree (an old phone number, a previous address, "St." on one and "Street" on another plus a different suite), the assistant has to guess, and it often picks the version that appears most.
Write down one canonical version of your business name, address, phone, hours and website, then make every profile match it: Google Business Profile, Bing Places, Apple Business, Yelp, Facebook, and the directories for your trade.
07Check what Perplexity says about you, then re-check monthly
Ask Perplexity the questions your customers ask ("best emergency plumber in Leeds", "dentist open Saturdays near Austin"), not just your business name. Note who gets named, what's said about you, and which sources are cited. Perplexity shows every source it used, so click through the cited pages that outrank you and compare them with yours: they are usually more specific, more recent, or listed on a site you aren't.
Our free AI Visibility Quick Scan does a first pass, and this guide shows how to test by hand.
08Optional: add an llms.txt
llms.txt is a proposed plain-text summary of your site for AI tools. None of the major assistants has said it uses llms.txt to choose what to cite, so treat it as a cheap extra, not a fix. It takes ten minutes with our llms.txt Generator. More in Does llms.txt actually help?

Sources

We update this guide when Perplexity changes how it works. Something out of date? Tell us.

Free tools for this guide

No account. They only read public files.

FreeNo signupAbout 10 secondsPublic files only

FAQ

Perplexity: common questions

Other assistants

All engine guides →

Rather not do this yourself? We'll do the whole thing and send you the receipts.

About the audit

How does Perplexity find businesses?

Which crawlers matter for Perplexity?

What should you fix, step by step?

01Allow PerplexityBot, in robots.txt and in your firewall

02Answer specific questions on specific pages

03Put your key facts in plain HTML text

04Add LocalBusiness schema that matches your listings

05Earn real reviews where this engine looks

06Make your name, address and phone identical everywhere

07Check what Perplexity says about you, then re-check monthly

08Optional: add an llms.txt