How to Appear in Perplexity and Get Your Brand Cited
Perplexity shows its sources next to every answer, and it publishes which agents fetch them and from which addresses. What decides whether a site is among them, what a company can check today on its own domain, and how much of the work is already done if it invests in SEO.
Published 24 September 2026
Short answer
Appearing in Perplexity is the outcome of the assistant naming a brand or linking one of its pages inside the answer it writes, and it is earned by letting its two agents in and publishing pages that settle the question with attributable facts. Perplexity documents both agents: PerplexityBot, the one that gets a site surfaced and linked in its results, and Perplexity-User, which fetches a page when someone asks a question that needs it. Perplexity is also the assistant whose citations overlap most with Google’s first results: 28.6% of the URLs it cites sit in Google’s top 10 for the same query, against an average of 12% across the assistants Ahrefs measured. A company already working on its rankings therefore starts with part of the road behind it.
What is appearing in Perplexity
DefinitionAppearing in Perplexity is the outcome of the assistant naming a brand or linking one of its pages inside the answer it writes. It is not a position in a list of results and it is not a slot anyone can buy: Perplexity drafts an answer and, beside each claim, shows where the claim came from.
Perplexity reaches those sources two ways. In one it searches its own index, the one it builds by crawling the web on its own schedule. In the other it fetches a specific page on the spot, because the question someone just typed calls for it.
For a company the practical consequence is that there is nothing to apply for and nothing to pay. What exists is two agents that have to be able to reach the site, and pages that answer a concrete question with figures, dates and conditions someone can attribute.
Which Perplexity agents decide whether it can cite you
AgentsPerplexity documents two agents, and each one is allowed or blocked separately. One gets a site surfaced and linked in its results; the other fetches a page when someone asks a question that needs it.
| Agent | What Perplexity uses it for | What happens if it is blocked | Where it is controlled |
|---|---|---|---|
| PerplexityBot | Surfacing and linking websites in Perplexity search results; the documentation states it is not used to crawl content for AI foundation models | Perplexity’s documentation ties the site appearing in its results to that permission | The site’s robots.txt, plus the IP ranges Perplexity publishes for that agent |
| Perplexity-User | Visiting a page when someone asks a question, to answer with it and link to it in the reply | That page is not used to answer that particular question | The firewall or the CDN: Perplexity documents that this agent, because a person requested the fetch, generally ignores robots.txt rules |
Perplexity recommends in that same documentation allowing PerplexityBot in robots.txt and, on top of that, permitting the requests arriving from the IP ranges it publishes. Two permissions in two different places, and the second one sits outside the reach of whoever writes the content: it lives in the firewall.
A change to those rules is not instant either. Perplexity’s documentation warns that its systems may take up to 24 hours to reflect it, so a check run ten minutes later proves nothing.
What robots.txt can and cannot do about Perplexity
Real controlA robots.txt file states which agents may fetch a site, but it stops no fetch technically. Perplexity documents that its user-request agent generally ignores those rules, precisely because a person is waiting on the other end.
In August 2025 Cloudflare, the infrastructure company a large share of internet traffic passes through, published that it had observed fetches attributed to Perplexity rotating addresses and networks to get around blocks. Cloudflare de-listed Perplexity as a verified bot and added detections for that crawling to its managed rules. The post went up on 4 August 2025 and was updated on 15 July 2026.
The useful conclusion for a company does not depend on who is right in that disagreement. If a management team wants to genuinely decide who reaches its website, the decision belongs in the layer that can actually stop a request: the firewall or the CDN, where traffic is allowed or blocked by agent name and by source address.
Perplexity’s own documentation explains how to build that rule in Cloudflare and in AWS, and asks for both conditions together, agent name and address, rather than trusting the name alone.
Why the SEO you already did counts on Perplexity
OverlapPerplexity is the assistant whose citations overlap most with Google’s first results. 28.6% of the URLs it cites sit in Google’s top 10 for the same query, according to the analysis of 15,000 prompts Ahrefs published.
That share only makes sense next to the average. In the same study, on average just 12% of the URLs cited by AI assistants appear in Google’s top 10 for the original prompt, and more than 80% of their citations go to pages that do not rank for it at all. The study is by Louise Linehan, published on 11 August 2025 and updated on 31 May 2026.
For whoever signs off the budget, the reading is concrete: of all the assistants measured, this is the one where work already done for Google carries furthest. Every engine assembles its source list its own way, and how each surface picks what it cites is a subject of its own.
What the figure does not say is how long it takes to show. It measures the overlap between cited URLs and ranking URLs, not timelines, so it helps decide where to start rather than promise a date.
Which websites Perplexity cites most
SourcesReddit and YouTube take most of Perplexity’s citations. In the September 2026 snapshot of the Ahrefs index of most-cited domains, Reddit holds 21.6% of mentions and YouTube 20.8%.
| Domain | Mention share (September 2026) | What kind of source it is |
|---|---|---|
| reddit.com | 21.6% | Conversation between people |
| youtube.com | 20.8% | Video published by third parties and by brands themselves |
| en.wikipedia.org | 6.3% | Encyclopedia |
| facebook.com | 4.5% | Company profiles and pages |
| amazon.com | 3.6% | Catalogue and buyer reviews |
| healthline.com | 2.6% | Specialist publisher |
What that split tells a company is not that it should start posting on Reddit. It says a brand enters the answers two ways, through its own site and through sites it does not control, and that here video and conversation take up room they do not take up on a Google results page.
The split comes with two limits worth writing down. It measures United States queries across all topics, so a specific industry in another language can look very different, and Ahrefs recalculates the index every month, which makes it a snapshot rather than a constant.
How to appear in Perplexity step by step
The four stepsFour checks in this order, written to be handed to whoever runs the site. The first two take an afternoon; the last two are months of work.
- 01 / ACCESS
Allow PerplexityBot in robots.txt
The robots.txt file is where a site states which automated agents may fetch it. Perplexity recommends allowing PerplexityBot there so the site appears in its results. It is one line of text that the AI crawler test can check, signed off by whoever administers the domain, and it does not depend on the marketing team.
- 02 / FIREWALL
Allow the addresses Perplexity publishes as well
Perplexity publishes the addresses its agents come from in a file: the PerplexityBot one held eight ranges on 24 September 2026. Its documentation asks for the agent name and those addresses to be combined in the firewall, and warns the rule can take up to 24 hours to take effect.
- 03 / TEXT
Put the thing you want cited into text
Perplexity shows the source beside each claim, so what is quotable is what can be read and attributed: a figure with its date, a concrete condition, a definition. On many sites exactly that — prices, lead times, conditions — loads after the initial HTML, and what arrives later may not be in what the agent fetched.
- 04 / SOURCES
Be described accurately on the sources this engine already cites
Part of the answers leans on sites a company does not control: forums, video, encyclopedias and publishers. Getting those sources to describe the brand with the same name, the same category and the same facts as its own site is relationship work measured in months, and it is what holds up whatever the assistant says about it.
What is missing from the documentation is informative too. Perplexity’s crawler documentation asks for no new file and no special markup: it covers robots.txt and allowing its addresses in the firewall, and nothing else.
How to tell whether Perplexity already visits your site
MeasurementServer or firewall logs are the most direct source: they record every visit with the name of the agent that made it. That is where Perplexity’s two agents separate, and each one says something different.
A visit from PerplexityBot says the engine is crawling the site for its index. A visit from Perplexity-User says something more valuable: that someone asked a question whose answer led the assistant to that specific page.
Both are checked against the addresses Perplexity publishes, because an agent name can be written into any request, whereas the source address is matched against a list the engine itself maintains.
The other route is not technical, and it is the one that has produced usable data across DoGeo’s projects: asking each customer how they arrived. At HeyTalent, the leads who say so name mostly ChatGPT and Gemini, with Perplexity and Claude in a few cases, according to what HeyTalent’s leads state in sales conversations. It is what people say when they sign, not a measured split: there is no per-engine percentage behind it.
Keep reading
Want to know what Perplexity sees when it reaches your site?
DoGeo checks which agents can fetch your pages, how much of your content can be quoted as it stands, and which outside sources already name you. Four details are needed: your name, your company, your site and a work email.