How to get your business recommended by Perplexity

Perplexity builds nearly every answer from web pages it fetches at the moment you ask, so getting recommended there is mostly about having a page that answers the question and can be read without trouble. It is the most direct of the assistants in this sense. There is far less waiting for a model to learn who you are, because it is looking things up rather than remembering.

Why it behaves differently from the others

Perplexity shows its sources next to the answer as numbered references. That tells you two useful things. It is retrieving rather than recalling, and it is willing to show you exactly which pages it used. If you ask it a question in your category and your competitor appears with three citations, you can open all three and see what they published that you did not.

That makes it the best place to start measuring. You get a clear reading and a list of the pages that beat you, on the same screen. No other engine hands you the diagnosis and the result together.

Perplexity also publishes the names of its crawlers, which matters if your site blocks anything. PerplexityBot is the crawler that indexes pages, and Perplexity-User is the fetch made when a user's question causes a page to be opened. If your firewall or your robots.txt blocks these, you have removed yourself from the engine that would otherwise react fastest to your work. Check that before you write anything.

The five things that get a page picked

It answers the exact question. Not the topic, the question. A page titled for the question a buyer typed, with the answer in the opening lines, is far easier to lift than a services page that happens to mention the subject.

It can be read as plain text. If your content only appears after scripts run, the fetch may return almost nothing. This is the most common silent problem, and it is invisible when you check the page yourself in a browser.

It has something specific to quote. A count, a named list, a price with a date, a step by step method. Retrieval finds many pages that are broadly about the subject. The one that gets quoted is the one carrying a sentence worth repeating.

The claim is supported somewhere else too. When two sources agree, a summary that names you is safer to write. When only your own site says it, you will often be read and then left out.

It is not blocked. The dullest item on the list and the one that silently cancels the other four.

A worked example

Here is a real one, and it is the shortest version of the whole method. On 6 August 2026 ChatGPT cited the page clawlaw.in/blog/how-to-check-a-companys-court-cases-in-india as a source for a vendor due diligence question. That page had been published fourteen days earlier. At baseline, on the same question set, the company had been named nowhere in that part of the market. clawlaw.in is our sister company, and this is one of the readings a reader can check independently, because the page and the question are both public.

Two things in that reading are worth copying. The page was built for one question a buyer actually types, and it answered the question rather than describing a product. And the gap between publishing and being used as a source was fourteen days, not a quarter. On an engine like Perplexity, which fetches pages while it answers, fourteen days is the sort of timescale you are working on.

So the routine is: publish one page for one question, wait a few days, then ask that question here in a fresh conversation and open every citation. Read them the way the engine did. Roundups supply the list of candidates. A pricing page supplies the numbers that get repeated. Documentation answers the "does it work with" half of the question. A forum thread supplies the sentence about what goes wrong. A consultant's post supplies an apparently neutral opinion.

Then your work is specific rather than vague. Missing from a roundup is an email. No numbers on your pricing page is a decision more than a writing task. Documentation behind a login is a configuration change. The difficult question that only a forum answers is a page you can write honestly this week. Four jobs, all of them small, and each one aimed at a source that is demonstrably in use.

How to check where you stand

Ask it the questions your buyers ask, in their words, and never name your own company in the question. Then read the citations, not only the answer. Note which domains appear over and over across your questions, because that list is your real competitive set on this engine, and it usually includes a directory or a comparison site you are not on.

Record the citations by full URL rather than by domain. Two pages on the same review site can behave completely differently, and you will want to know which one to get onto.

What not to expect

Being retrieved is not the same as being recommended. Your page can be cited as a source while the answer recommends somebody else, and that is still progress worth tracking separately.

Retrieval also does not fix a weak page. A page that is fetched and has nothing specific in it gets read and skipped, which looks identical to never being found and needs the opposite fix.

And the citation list shows what was used, not everything considered. A page can be fetched, read and left out. So absence from the citations is weaker evidence than presence in them, and no one can promise you a place in an answer, because the same question asked twice can come back differently.

Common questions

How often should I check my position here?

Weekly is reasonable on this engine, and it is the one place where weekly is not a waste. Because it retrieves live, a new page can show up within days, so a weekly reading of your top questions genuinely tells you whether publishing worked. Run the full set monthly.

Why do two runs of the same question give different companies?

Because each run does a fresh search and can retrieve a different set of pages. The search itself varies, results move, and the mode you use changes the depth of retrieval. Treat any one answer as a sample. If you need a number, take several readings and report the range rather than the middle.

Perplexity cited a competitor's page for a fact about my company. How do I fix that?

This is the most fixable version of the problem, because you can see the page. Open it. If the fact is wrong, write to the owner with the correct value and a link to your own page stating it. If the fact is right but it is their page rather than yours, that is a gap in your own site: publish that fact where a buyer asking that question would land, in text, with a date, and make sure the page is fetchable.

Does being cited mean I am being recommended?

No, and here the difference is unusually visible because the citations are numbered. You can be citation four in an answer that recommends two other companies. Keep both columns in your sheet. Citation means the mechanics work. A mention in the text means the content persuaded.

Should I write pages aimed at Perplexity specifically?

No. Write pages aimed at one buyer question each, with the answer at the top and something checkable in them. That is the same page that works on the other engines. What is specific to Perplexity is the measurement, not the writing.

Do I need third party reviews to be recommended?

They help when they exist, because a claim supported by a source you do not own is safer to repeat. But in some categories, particularly newer or narrower ones, independent reviews barely exist. Do not wait for them. In those categories the winning route is to be the vendor whose own pages are the easiest to verify.

What to do first

Before you write a word, confirm you are not blocked. Check your robots.txt and any firewall rules for PerplexityBot and Perplexity-User, and fetch one of your own pages in a way that shows you the raw HTML rather than the rendered view. If your content is not in that HTML, fix that first, because nothing else on this page can work until it is.

How we know this

We read the assistants rather than guessing what they do. In our own audit of 1,260 saved answers from ChatGPT and Gemini across the United States, the United Kingdom and India, 82.0% of every citation landed on a site the business itself owned, and review directories took 0.7%. Counted as one domain per reading, 16,516 of 20,143.

That is the whole asymmetry: the pages you own are the pages an assistant quotes back to a buyer. See the dated readings and the method on our proof page.

Find out whether assistants name your business, on your own domain and your own buying questions.

By Aditya Goel, Founder, AI Knows Us. Reviewed 1 November 2026.