Does blocking PerplexityBot protect my content from AI training?

No, because Perplexity states that neither of its crawlers is used to collect content for training foundation models. Blocking PerplexityBot removes you from Perplexity's search results without withholding anything from a training corpus, so it is a cost with no matching benefit.

Why does Perplexity fetch my pages even though robots.txt disallows it?

Perplexity-User is the likely explanation. Its documentation says that because a person requested the fetch, it generally ignores robots.txt rules. That behaviour is documented rather than accidental, and it means robots.txt is not a reliable control for user-initiated retrieval.

I allow the crawlers but see no Perplexity traffic. What now?

Check your firewall before your robots file. A WAF or CDN bot rule will challenge or drop an unfamiliar agent before robots.txt is consulted. Perplexity publishes IP ranges for exactly this reason, and allowlisting them is more reliable than matching on a user-agent string.

Can Maveriko tell me whether Perplexity has cited my site?

No. Maveriko reads your page and your public entity records, so it can tell you whether the crawlers are allowed and whether the content is structured to be quoted. Whether an engine actually quoted you is an output that only repeated live prompting can observe.

Perplexity SEO

Perplexity SEO is the work of making a page reachable and quotable by Perplexity's two crawlers. Both exist to find and link sources, and Perplexity's own documentation states that neither is used to collect content for training foundation models.

That makes Perplexity the simplest of the answer engines to reason about. There is no trade-off to weigh between feeding a training corpus and appearing in live answers, because the training question does not arise. If you want to be cited, you allow the crawlers. If you block them, you are not in the index they answer from.

Score any live URL — free, no account

The two crawlers, and what each one does

Perplexity publishes two user agents with different jobs, and the distinction matters when you write a robots.txt rule.

The second row carries a caveat worth reading twice. Perplexity documents that because a person asked for the fetch, Perplexity-User generally ignores robots.txt. A disallow rule is therefore not a reliable way to keep a page out of a user-initiated answer, and treating it as one gives you a false sense of control.

User agent	What it does	Trains a model?
PerplexityBot	Surfaces and links websites in Perplexity search results. Allow it if you want to appear.	No
Perplexity-User	Fetches a page when someone's question causes Perplexity to open it, then links it in the answer.	No

Robots.txt is not the only gate

A crawler can be welcome in robots.txt and still be turned away at the door. Web application firewalls block unfamiliar automated traffic by default, and that block happens before your robots rules are ever read.

Perplexity publishes its IP ranges specifically so you can allowlist them, and gives configuration guidance for Cloudflare and AWS. If your analytics show no Perplexity traffic while your robots.txt plainly allows it, the firewall is the first place to look, not the second.

Confirm robots.txt does not disallow PerplexityBot — a blanket rule aimed at scrapers usually catches it.
Check your WAF or CDN for bot-management rules that challenge or drop unknown agents.
Allowlist the published IP ranges rather than trusting the user-agent string alone, which anyone can forge.
Re-scan the page afterwards and confirm the crawler-access check passes.

What to fix on the page itself

Access only gets the page read. Being quoted depends on whether there is a clean, self-contained claim to lift, and that is the same work every answer engine rewards.

The highest-yield changes are unglamorous: open each section with a sentence that answers its own heading, keep that sentence short, mark up what the page is with schema, and put enumerable facts in a list or a table rather than burying them in prose.

A direct-answer opening under each heading — a definition or a fact, not a wind-up.
Structured data that declares what the page is about, not only who published it.
Lists and tables where the content is genuinely enumerable, because a table row is the easiest thing in HTML to quote.
Visible publication and review dates, so an engine can judge whether the claim is current.
Outbound citations to primary sources, which is the signal that a claim is checkable.

What none of this can tell you

No tool reads your HTML and reports how often Perplexity actually named you. That requires running real prompts against a live model repeatedly, and the answers move as the model and the index change.

What is measurable from your page is readiness: whether the crawlers are permitted, whether the markup declares the subject, whether the prose is shaped to be lifted. Those are the inputs you control, and they are the honest thing to score.

Frequently asked questions

Does blocking PerplexityBot protect my content from AI training?: No, because Perplexity states that neither of its crawlers is used to collect content for training foundation models. Blocking PerplexityBot removes you from Perplexity's search results without withholding anything from a training corpus, so it is a cost with no matching benefit.
Why does Perplexity fetch my pages even though robots.txt disallows it?: Perplexity-User is the likely explanation. Its documentation says that because a person requested the fetch, it generally ignores robots.txt rules. That behaviour is documented rather than accidental, and it means robots.txt is not a reliable control for user-initiated retrieval.
I allow the crawlers but see no Perplexity traffic. What now?: Check your firewall before your robots file. A WAF or CDN bot rule will challenge or drop an unfamiliar agent before robots.txt is consulted. Perplexity publishes IP ranges for exactly this reason, and allowlisting them is more reliable than matching on a user-agent string.
Can Maveriko tell me whether Perplexity has cited my site?: No. Maveriko reads your page and your public entity records, so it can tell you whether the crawlers are allowed and whether the content is structured to be quoted. Whether an engine actually quoted you is an output that only repeated live prompting can observe.

Sources

Last reviewed 7 Sept 2026.

Fixing one page? Audit the whole site.

These tools work on a single page, in your browser. The full live-URL audit — SEO, GEO and entity authority — runs free with no sign-up and hands you your top fix; sign in with Google (also free) for the rest of the fixes and your saved report history.