Not Cited by Perplexity? Test PerplexityBot Access, Fix It

Using the sources list as evidence

Unlike assistants that sometimes answer without links, Perplexity is built around showing where an answer came from. Each reply carries numbered references and a sources view listing the pages it read. That list is your first diagnostic tool, because it tells you not just that you are missing but who took your place and what kind of page they offered.

Do this check in a logged-out or fresh session, and run each question two or three times. Perplexity's sources vary between runs, so a single missing appearance means little; a domain that never appears across a dozen runs is a real signal.

Perplexity's two user agents, as documented

Perplexity publishes a short page describing its crawlers. Read it before changing anything, because it is the authority on names and behaviour and it can change. At the time of writing it describes two agents with quite different jobs:

Perplexity also publishes the IP address ranges both agents use, as JSON files linked from that documentation, and recommends that sites which want to be included allow both the user agent and those ranges through their firewall.

The split between the two agents is what makes the tests in this guide work. If PerplexityBot is blocked in robots.txt but nothing else is wrong, Perplexity-User can often still fetch a page when you ask about it directly. If both fail, the block is almost always at the firewall, CDN or server, because those layers do not care what robots.txt says.

The direct URL test

This is the quickest single test for Perplexity and it needs no server access. Pick a page with distinctive content, such as a price list or a staff page, and ask Perplexity something only that page can answer, including the full address:

Summarise https://www.example.co.uk/services/boiler-servicing
and tell me the price it gives for an annual service.

Then read the reply carefully:

• It quotes the right price and lists your page as a source. The page can be fetched and read. If you still never appear for normal questions, the problem is either PerplexityBot being blocked in robots.txt, discovery, or competitiveness. Check robots.txt next.
• It says it cannot access the page, or summarises other sites instead. The fetch failed. Because Perplexity-User generally ignores robots.txt, a failure here points to a firewall, bot protection, geo-blocking, a login or a server error.
• It fetches the page but gets the content wrong or thin. The fetch worked but the HTML it received does not contain the text, usually because the content is loaded by JavaScript, hidden behind a consent screen, or inside an image.

Record the exact time you ran the test. You can then find that request in the access log in the next step and see the status code your server gave it.

Confirming it in your logs

The direct URL test tells you something failed; the log tells you what. Search around the time of the test for both agents and look at the status codes:

# Every Perplexity request, with status and path
grep -E "PerplexityBot|Perplexity-User" access.log \
  | awk '{print $1, $9, $7}' | sort | uniq -c | sort -rn | head -40

The first field is the client IP address. Compare a few of them with the ranges in Perplexity's published JSON files. Genuine requests from those ranges receiving 403, 429 or 503 responses are your block. Requests that claim to be PerplexityBot from addresses outside those ranges are impostors and can be treated like any other unknown bot.

If the log shows no Perplexity request at all at the moment you ran the direct URL test, the request never reached your server. Look at the event log of your CDN or hosting firewall for that minute; the block will be recorded there, often with the name of the rule that fired.

A short Python check saves comparing addresses by hand. Download the JSON file named in Perplexity's documentation and test an address from your log:

import ipaddress, json
ranges = json.load(open("perplexitybot.json"))
nets = [ipaddress.ip_network(p.get("ipv4Prefix") or p.get("ipv6Prefix"))
        for p in ranges["prefixes"]]
ip = ipaddress.ip_address("203.0.113.25")  # replace with the address from your log
print(any(ip in n for n in nets))

The keys follow the layout that Google and OpenAI use for their own range files. Open the file you downloaded and adjust the names if Perplexity's layout differs.

Where Perplexity gets blocked, by platform

Pay particular attention to "block AI bots" lists pasted into .htaccess or server configuration. They circulate widely, often include both Perplexity agents alongside training crawlers, and are easy to forget once added.

# .htaccess: an inherited rule that silently blocks Perplexity
RewriteCond %{HTTP_USER_AGENT} (GPTBot|PerplexityBot|Perplexity-User|ClaudeBot) [NC]
RewriteRule .* - [F,L]

If you want to keep blocking training crawlers, edit the list to remove only the agents that you want to answer questions about your business, rather than deleting the whole rule.

The fix, step by step

Unusual causes worth ruling out

• A copy of your content wins. If a partner republishes your articles, or a scraper copies them, Perplexity may cite the copy. Ask legitimate partners to use a canonical link to your original and to publish after you; report scrapers to their host. Make sure your own page is the fastest to load and the clearest to read.
• You recently moved domain or restructured URLs. Old addresses that redirect through several hops, or that return 404, lose their place. Keep redirects to a single hop and update your sitemap so the new addresses are discovered.
• Your server is slow or times out under load. A fetch that takes too long may be abandoned. Check the log for requests from Perplexity that took several seconds or ended in 5xx errors, and fix the performance problem.
• A shared name. If another business uses a similar name, Perplexity may be finding the other one. Include your town, trade and web address consistently wherever you are listed.
• You have chosen to block it. Blocking Perplexity is a legitimate choice. Just be aware that blocking PerplexityBot in robots.txt alone may not stop user-requested fetches, according to Perplexity's own description, so a firewall rule is the only complete block.

Keeping Perplexity access working

Perplexity access tends to break for the same reason it broke the first time: a security tool or a copied rule. Add a monthly check to whoever looks after the site:

1. Run the direct URL test on one key page and confirm it quotes the correct detail.
2. Count PerplexityBot requests and their status codes in the last month's log.
3. Re-download Perplexity's IP ranges if you allow-list by address, because ranges change.
4. Read the robots.txt served in production after any deployment or plugin update.
5. Rerun a fixed question set in the AI Visibility Tracker and compare with last month.

The same checks for every AI assistant, rather than Perplexity alone, are in fixing robots.txt AI blocks.