Perplexity is the friendliest surface for a product nobody knows yet

If you are going to spend a week understanding one AI surface properly, and you have a product that launched recently, this is the one that will repay it fastest.

Run the free check

Perplexity composes answers primarily from current retrieval with visible citations, which means a page that answers a question well can be cited without the domain having any history. Perplexity documents two crawlers, PerplexityBot for surfacing and linking sites in search results and Perplexity-User for user-initiated visits, and states that neither is used to crawl content for training foundation models.

Why this surface suits a new product

Perplexity's answers are built around retrieval with visible citations. The sources are listed, numbered, and clickable, and the answer is composed from them.

That structure has a direct consequence for a new product. When the answer is assembled largely from what was just retrieved, the question of whether the model already knew about you matters less than it does elsewhere. A page that answers the question well is a candidate.

The visible citation also means the traffic actually arrives. An answer that summarises your page without naming it gives you nothing. An answer that lists your page as source three sends people who click it.

The crawlers, and the decision you do not have to make

Perplexity documents two agents. PerplexityBot is described as designed to surface and link websites in search results on Perplexity, and its documentation states it is not used to crawl content for AI foundation models. Perplexity-User handles user-initiated requests, visiting pages when somebody asks a question so the response can cite sources, and likewise does not gather data for training.

This makes the robots.txt decision simpler here than for other companies. With OpenAI and Anthropic you have a genuine choice to make about training crawlers, and reasonable people land differently on it.

On Perplexity's own documentation there is no training crawler to weigh, so blocking these agents means declining citations and getting nothing in return. If your robots.txt currently blocks them, that was almost certainly collateral damage from a blanket AI block rather than a decision anybody made.

What gets cited

The same passage-level qualities that matter across the AI channel, applied here with unusual visibility because you can see the sources listed and go and read them.

That is the most useful habit available on this surface. Ask a question your product should answer, look at which sources were cited, open every one, and read them as a competitor would. You will see quickly what shape they share.

Usually it is the same shape: a page devoted to that specific question, answering it near the top, in plain language, with concrete specifics rather than adjectives, on a site that is readable without executing script.

Follow-up questions are where you can win

Perplexity suggests related questions after an answer, and people follow them. This creates a chain of increasingly specific questions, and specific questions are exactly where a small focused product can compete.

The broad opening question in your category is contested by everybody. The third follow-up down, which asks about a particular constraint or edge case, is often contested by almost nobody, because established players write for the broad question and stop.

So the practical exercise is to run the chain yourself. Ask the broad question, follow the suggested questions three or four levels deep, and note every one where the answer is thin or the sources are poor. Each of those is a page worth writing, and each of them is winnable now rather than in a year.

Doing the measurement honestly

Fix a list of questions your product should be cited for, ask them monthly, and record which sources were cited and whether you were among them.

Keep the questions identical between runs, because changing them destroys the comparison, which is the whole value. Add new ones if you like and leave the originals alone.

Also check your analytics for perplexity.ai as a referrer. If you have never seen it, either the citations are not happening or your analytics is filing the traffic somewhere else, and both are worth knowing about. Referral traffic from AI surfaces is frequently misfiled as direct.

What to do in the first week

Concrete, in order, and all of it fits into a few evenings.

  • Read your robots.txt and confirm PerplexityBot and Perplexity-User are not blocked. On Perplexity's documentation there is no training trade-off to weigh, so a block here costs you and returns nothing.
  • Ask the five questions your product should be the answer to, and record which sources were cited for each one.
  • Open every cited source and read it as a competitor would. Note where the answer sits on the page, how specific it is, and whether the page is devoted to that one question.
  • Follow the suggested related questions three or four levels deep and write down every one where the answer is thin. Those are the pages worth writing, and they are winnable now.
  • Pick the single thinnest one and write that page this week, with the answer in the first sentence under a heading that is the question itself.

Keeping it in proportion

Perplexity is smaller than Google and smaller than ChatGPT, so this is not the channel that carries a business.

Its value for a new product is different: it is the surface where the gap between what you deserve and what you get is smallest, which makes it the best place to learn whether your content is actually any good. If a page cannot get cited on Perplexity, where domain history matters least, the problem is likely the page rather than your circumstances.

Treat it as a proving ground as much as a traffic source. The work that earns a citation here is the same work that eventually earns one elsewhere.

Part of a larger guide

This page is one part of The AI channel. The other parts:

Questions people ask

Why is Perplexity easier for a new website?
Because its answers are composed largely from current retrieval with visible numbered citations, so whether the model already knew about your domain matters less than it does on other surfaces. A page that answers the question well is a candidate, and a visible citation means the traffic actually arrives when somebody clicks it.
Should I allow PerplexityBot?
Yes, on Perplexity's own documentation there is no training trade-off to weigh. It describes PerplexityBot as designed to surface and link websites in its search results and states it is not used to crawl content for AI foundation models, so blocking it declines citations and gets nothing back.
What is the difference between PerplexityBot and Perplexity-User?
PerplexityBot surfaces and links websites in Perplexity search results. Perplexity-User handles user-initiated requests, visiting pages when somebody asks a question so the response can cite sources. Perplexity's documentation states that neither is used to gather data for training foundation models.
How do I find topics I can win on Perplexity?
Ask the broad question in your category, then follow the suggested related questions three or four levels deep. The broad question is contested by everybody, but the specific follow-ups are often thin, because established players write for the broad question and stop. Each thin answer is a page worth writing.
How do I track whether Perplexity is citing me?
Fix a list of questions your product should be cited for, ask them monthly without changing the wording, and record which sources appear. Also check your analytics for perplexity.ai as a referrer, keeping in mind that AI referral traffic is frequently misfiled as direct.
Is Perplexity worth the effort given its size?
It will not carry a business, being smaller than both Google and ChatGPT. Its value is as a proving ground: it is the surface where domain history matters least, so if a page cannot earn a citation there, the problem is likely the page rather than your circumstances.

Check that Perplexity can reach you

The free Get Found Check reports whether PerplexityBot and the other search agents are currently allowed on your domain, and whether your pages are readable when they arrive.

Run the free check Have Licheo do it for you