How to appear in Perplexity
Perplexity always searches the live web and always cites. YouTube at 32.4% of citations, communities carrying weight, and 52% of citations linking to the source without naming the brand. What to do to be read, and to be said.
Open in ChatGPT
Open in ClaudePerplexity is the engine closest to SEO in speed and the furthest in sources: it always searches, it always cites, and what it cites comes from YouTube and from communities before it comes from your site.
How Perplexity decides
It always searches the live web. There is no memory path: the question triggers a search, the engine reads pages at that moment and answers citing them. Whoever has findable, fast content that answers the question in the right passage wins.
With YouTube and the communities in front. YouTube leads citations at 32.4% (Ahrefs, June 2026); Reddit reaches around 47% in some category studies (directional); primary research and specialist press weigh more here than in the other engines (5W, June 2026).
No advertising in the answers. By public decision. Earned presence is the only path.
- 32.4%of citations go to YouTube, the leading source
- 47%of citations go to Reddit in some category studies
- 52%of citations link to the source without naming the brand
The Reddit figure is approximate (around 47%), directional, and varies by sector. Sources: Ahrefs, June 2026 (YouTube); category studies (Reddit); Writesonic via Search Engine Land, July 2026.
What to do, in order
Allow PerplexityBot and have fast pages
Perplexity always searches the live web. There is no memory path to make up for it: if the page is not read at that moment, it does not enter the answer. PerplexityBot feeds the index; Perplexity-User fetches the page when somebody asks for it in a conversation. A slow page, or one that only exists after the JavaScript runs, is a page the engine gives up on reading.
Do: in robots.txt, allow PerplexityBot and Perplexity-User. Server-rendered HTML, low response time, and the answer to the question in the first sentence of the right section.
Be in the communities, and get mentioned in them
Communities weigh heavily in this engine: Reddit reaches around 47% of citations in some category studies (directional, varies by sector). It is not a publishing channel: it is third-party conversation, and it is won by deserving it. A brand that only appears in threads through the hand of whoever sells it is thrown out by the community and learned by the engine.
Do: find out in which threads your category is discussed, take part with useful, signed answers, and treat the category forums as press: it is there that Perplexity goes for the opinion it cites.
Publish video on YouTube and data worth citing
YouTube leads Perplexity's citations, at 32.4% (Ahrefs, June 2026). Primary research portals and specialist press weigh more here than in the other engines (5W Citation Source Index, June 2026; vendor data, treat as directional). Original data, published with method, is the raw material those sources cite, and the engine behind them.
Do: one video per buyer question that matters, and one original study per quarter, however small, with a published method and a downloadable file.
Write the brand name inside the citable passage
More than half of Perplexity's citations link to the source without naming the brand in the text of the answer: 52%, according to Writesonic (via Search Engine Land, July 2026). That is, the page is read and linked, and the buyer never sees the name. The cause is on the page: the passage the engine extracts does not have the brand inside it.
Do: in the sections that answer questions, write the brand name in the sentence that answers, and not only in the heading or the footer. It is the difference between being a source and being a recommendation.
Measure every week, because here it changes faster
Because it always searches, Perplexity is the engine where the answer changes fastest, for better and for worse. The page it reads today may not be the one it reads next week. It is also the engine where the work shows up first: weeks, not months.
Do: a fixed set of real buyer questions, run every week, with the cited sources kept. It is what we do, with Perplexity always in search mode because it is the only one it has.
The check, in one command
What is verified with a command. Swap the domain for yours.
The Perplexity agents have a way in
$ curl -s https://example.com/robots.txt | grep -iA1 'PerplexityBot\|Perplexity-User' User-Agent: PerplexityBot Allow: / -- User-Agent: Perplexity-User Allow: /
Two agents, two blocks. If you only have a generic rule with Disallow, the two are left out and, since there is no memory path, the brand does not exist in this engine. The full check is in six commands to find out whether the AI can read your site.
What does not work
Publishing in the communities as if they were a brand channel: the communities throw you out and the engine learns. Expecting memory to make up for a slow page: there is no memory. And having the brand only in the heading: the passage the engine extracts is what the buyer sees, and in more than half of the citations it does not carry the name.
Write the brand name in the sentence that answers, and not only in the heading or the footer. It is the difference between being a source and being a recommendation.
Frequently asked questions
No: it always searches the live web and cites explicit sources. That is the structural difference from ChatGPT and Claude, which mix memory with search. Without a page read at that moment there is no presence, and with a page read the presence can arrive in weeks.
No. Perplexity decided publicly not to sell advertising in the answers. Earned presence is the only path, and the sources it cites have names: YouTube, communities, primary research and specialist press.
Because the passage it extracted does not have the brand name inside it. It happens in 52% of citations (Writesonic via Search Engine Land, July 2026). It is fixed on the page: the brand name in the sentence that answers the question, and not only in the heading.
Weeks, because there is only the search path: as soon as the page exists, is fast, is read and answers the question, it can be cited in the next answer. It is also the engine where presence is lost fastest, for the same reason. Nobody serious promises a deadline; what is promised is weekly measurement.
Read next
- How to appear in ChatGPT, the engine that reads the Bing index.
- How to appear in Gemini, the other engine where YouTube leads.
- How to measure GEO, citation and share of voice as a series.