How to track brand mentions in Perplexity
Perplexity puts numbered source links beside its answers, which makes it the easiest AI search engine to audit by hand. The work is in doing it consistently: the same questions, a clean session, and a record of which pages Perplexity found, which it cited and which it passed over.
The short answer
- To see mentions today: ask buyer questions in an incognito thread, where Perplexity turns memory and previous searches off, and read the numbered citations and the source list.
- To track them over time: run the same questions weekly through Perplexity's API with web search, set your market with
user_location, and keep the search results it returns as well as the answer. - Keep the door open: allow
PerplexityBot, which Perplexity uses to surface and link websites in its results.
How Perplexity reaches your pages
Perplexity documents two agents, and they behave differently:
| Agent | What Perplexity says it does | robots.txt | IP list |
|---|---|---|---|
PerplexityBot | Surfaces and links websites in Perplexity search results; not used to crawl content for AI foundation models | Perplexity recommends allowing it | https://www.perplexity.com/perplexitybot.json |
Perplexity-User | Fetches a page when a user's question needs it, and may link to it in the answer | Generally ignores robots.txt, because a user requested the fetch | https://www.perplexity.com/perplexity-user.json |
Because PerplexityBot is the agent behind Perplexity's search results, blocking it in robots.txt or at your firewall works against being cited. Perplexity-User explains why a page can still be fetched after you block the crawler: someone asked about it directly.
Step 1: Check by hand
- Open an incognito thread. Perplexity's help center says memory and previous searches are always off in incognito mode, and those questions are not used in memory.
- Ask each question from your prompt set exactly as written: discovery questions that do not name you, comparisons with a competitor, and fact checks about your offer.
- Read the numbered citations in the answer, then the full list of sources, and note which of your URLs appear and in what order.
- Write down the search mode and model you used. Deeper research modes run more searches and can cite different pages, so compare like with like.
- Record the date, prompt, mode, whether you were mentioned and where, your cited URLs and the competitors named.
If you check from your own account instead, remember that Perplexity's memory draws on saved memories and your search history. The two can be switched off separately in the Personalize settings, and switching memory off does not delete what is already saved.
Step 2: Track it with the API
Perplexity's Agent API has a web_search tool. Its response includes a search_results item with:
queries, the searches it ran for your question;results, each with anid,url,title,snippet,dateandlast_updated.
Perplexity's documentation says the id and url of each result are the source of truth for citations. Inline citation markers such as [1] appear only if your prompt asks for them, so store the result list rather than parsing the answer text.
Two settings matter for brand tracking:
- Location:
user_locationtakes a two-lettercountrycode, plus optional region and city, so you can sample each market you sell in. - Domain filter:
search_domain_filterlimits a search to up to 20 domains, or excludes them. Do not use it while measuring, because it changes what Perplexity can find. Use it as a diagnostic instead: restrict one search to your own domain to see whether Perplexity's search returns any of your pages for that question at all.
Keep the model, settings and wording fixed between runs. API answers will not match the app exactly, because the app adds memory, search modes and its own settings.
Step 3: Read the results
| What you see | What it usually means | What to check first |
|---|---|---|
| Cited with a numbered link | Your page supports a claim in the answer | Keep facts and dates on that page current |
| Mentioned, not cited | Perplexity knows the brand from other pages | Which pages it cited instead |
| In the search results, not cited | Your page was found but not used | How directly and specifically it answers the question |
| Not in the search results | Your page was not retrieved | PerplexityBot access, then whether any page targets the question |
| Nothing even with your domain filter | Perplexity's search has no page of yours for this question | Whether such a page exists and can be crawled |
The date and last_updated fields let you see how recent the cited pages are compared with yours. Treat that as a clue to investigate, not a rule. Report mention rates with their counts, such as "cited in 6 of 24 answers", and wait for a change to persist across several runs before acting on it.
Fix access before content
Start with robots.txt:
User-agent: PerplexityBot
Allow: /Then check the layers beneath it:
- Firewalls and CDNs. Perplexity recommends matching both the user-agent string and its published IP ranges, and refreshing those ranges automatically rather than hard-coding them. Our Cloudflare AI Crawl Control guide covers the most common setup.
- Perplexity-User. It generally ignores robots.txt. Blocking it at the firewall stops Perplexity reading your page when a person asks about it, so make that choice deliberately.
- The page itself. Run the free AI visibility checker to catch noindex, canonical and rendering problems, and see the AI crawler list for the other engines' agents.
Measure visits from Perplexity
In GA4, look for sessions whose source contains perplexity. Treat the number as a minimum, because visits from apps can arrive without a referrer. In your server logs, Perplexity-User requests show pages that Perplexity fetched on demand for a person's question.
CiteVisor's paid plans sample Perplexity alongside ChatGPT, Claude and Gemini every week and keep each answer and source list. For the shared method, see the overview of tracking brand mentions in AI search; for the other engines, see the guides for ChatGPT and Gemini.
Related guides
- Track brand mentions in AI search, then compare ChatGPT and Gemini on the same prompt set.
- How to get cited by Perplexity turns missing sources into an access and content checklist.
- How to get cited by ChatGPT covers the other engine's retrieval path.
- Check Cloudflare AI Crawl Control and the AI crawler list, then run the free AI visibility checker.
Sources
- Perplexity: Perplexity crawlers
- Perplexity: Web search tool in the Agent API
- Perplexity Help Center: Memory
- Perplexity IP list for PerplexityBot
Reviewed on 2026-10-21. Perplexity changes its modes and settings often; check the linked pages before relying on a detail.
Frequently asked questions
How do I see mentions of my brand in Perplexity?
Ask the questions your buyers ask in an incognito thread, where Perplexity turns memory and previous searches off, then read the numbered citations and the source list. Record which of your pages appear and which competitors are named, and repeat on a schedule.
Does Perplexity use my website to train AI models?
Perplexity says PerplexityBot is not used to crawl content for AI foundation models, and that Perplexity-User is not used for crawling or for training. PerplexityBot is the agent that surfaces and links websites in Perplexity's search results.
Why does Perplexity cite a competitor instead of my page?
Check access first: if PerplexityBot is blocked, your pages are unlikely to be among its results. If your pages are retrieved but not cited, compare them with the cited page for that question: how directly it answers, how specific it is and how recently it was updated.
Can I stop Perplexity-User with robots.txt?
Not reliably. Perplexity says the fetcher generally ignores robots.txt because a user requested the fetch. A firewall rule can block it, but then Perplexity cannot read your page when someone asks about it.
Are there tools to track Perplexity mentions continuously?
Yes. Trackers sample Perplexity's API on a schedule; CiteVisor's paid plans do this weekly. Whichever you use, check that it keeps the raw answers and source lists so you can verify a change yourself.
Does blocking PerplexityBot stop citations?
Blocking PerplexityBot works against inclusion in Perplexity search results. It is not a guarantee that every reference to your site disappears: Perplexity-User can fetch pages for a user request and generally ignores robots.txt. Search crawling and user-triggered access are separate controls.
Perplexity brand mention tracker—what are the free options?
Use manual checks in an incognito thread and save each answer, its numbered sources, the prompt and the date in a spreadsheet. Repeat the same questions and mode weekly. This measures your sample; it is not continuous coverage of every Perplexity answer. API automation is paid separately.