Ana Fernández / SEO

What is Perplexity: how it works, pricing and citations

What Perplexity is, how it works, what Pro and Max cost, how it differs from Google and ChatGPT, and how it picks the sources it cites (PerplexityBot).

November 16, 2026 9 min readby Ana Fernández

What is Perplexity? It's an AI search engine that answers what you ask with a written response and numbered sources next to it, so you can check where each fact came from. Chances are someone on your team already uses it for research (and someone else has asked you whether your brand shows up there).

This guide covers both sides. First what it is, how it works, what it costs and how to use it. Then what happens on the other side, when your site is the source for that AI. My short take is that Perplexity is the AI search engine that looks the most like Google when it chooses what to cite, so showing up depends a lot on the same things as SEO, plus a couple of decisions about its bots.

What is Perplexity?

Perplexity calls itself an "answer engine". Instead of giving you ten blue links, it reads several pages and hands you direct answers, in a conversational tone, with numbered citations that lead to each source. You write to it the way you'd talk to a colleague ("which travel insurance makes sense for a month in Europe?") rather than as a telegram of keywords.

Perplexity was born in San Francisco. Aravind Srinivas, its CEO, founded it in 2022 with Denis Yarats, Johnny Ho and Andy Konwinski, with the idea of blending a traditional search engine with a conversational AI assistant.

Two details set Perplexity apart from any other chatbot. It searches the internet in real time every time you ask something, and it shows the sources above the text instead of hiding them. That made it popular for research, and it also made it interesting for those of us who publish content, because every answer leaves a visible trail of which sites it used.

How does Perplexity work?

Under the hood, Perplexity does three things in order. First, a language model interprets what you wrote and decides what to search for (sometimes it splits it into several smaller searches). Second, it searches its own index of pages and, when needed, visits sites in real time to bring in fresh information. Third, another model writes the result with what it found and adds a citation to each claim, so you can click and verify.

If you then ask a follow-up ("and for two people?"), it keeps the context of the conversation and you don't have to repeat everything.

Pro Search mode runs the same process with more steps and checks more sources before answering. According to the official help center, the free version allows 3 Pro Searches a day and subscribers get more than 300 daily.

If this sounds like RAG, that's because it is: it retrieves documents and generates the answers on top of them (I explained it calmly in what is RAG).

Which AI models does it use?

By default it runs on the "Best" option, which picks the model based on what you ask. Perplexity has its own model, Sonar, and on paid plans you can choose AI models from other companies, such as OpenAI's GPT, Anthropic's Claude or Google's Gemini. For everyday use the difference is barely noticeable; where it does show is in long questions or ones that need a lot of reasoning.

What is Perplexity used for?

Perplexity works for any search where you want a ready-made answer rather than a list of links to go through one by one. Some uses I see often:

  • Researching a new topic with the citations at hand.
  • Checking information that changes often, like a regulation, a price or this week's news.
  • Summarizing long documents: you upload a PDF and ask for the main points.
  • Quick market research, for example what the media says about a competitor or which options exist in a category.
  • Auditing your brand: asking it the questions a customer would ask, like which company it recommends for what you sell, and checking whether you show up, how it describes you and where it gets that from.

That last one is the one I care about most, and I'll come back to it at the end.

Is Perplexity free? Pro, Max and pricing

Yes, Perplexity has a free version that works for normal searches, even without creating an account. Paid plans add more advanced searches, model choice and more files.

PlanApproximate priceWhat it includes
FreeUS$0Basic searches with no practical limit, 3 Pro Searches a day, a few files a day
ProUS$20 a monthMore than 300 advanced searches a day, model choice, file uploads with no practical limit, Research and part of Labs
MaxUS$200 a monthEverything above, the newest models first, unlimited Labs and priority support

A practical heads-up: prices change depending on whether you pay monthly or yearly (Xataka, for example, reports US$17 and US$167), so check the pricing page before you hand over your card.

How to use Perplexity

Getting started takes a minute:

  1. Go to perplexity.ai in your browser or download the iOS or Android app. For simple queries you don't need to sign up.

  2. Write a specific question. "Mediterranean diet" gets you something generic; "what are the benefits of the Mediterranean diet for cholesterol?" gets you something useful.

  3. Use Focus if you want to limit where the information comes from, for example only academic papers or social media.

  4. Upload files (PDFs, spreadsheets, images) and ask it to analyze them. The free version allows only a few a day.

  5. Keep asking in the same conversation and save what matters in Spaces to pick it up later.

One tip for using Perplexity well: open at least two of the cited links before copying a fact. It takes ten seconds and saves you surprises.

What else does it do besides search?

In recent years Perplexity has added products around the search engine.

  • Research (formerly Deep Research) runs dozens of automatic searches, reads many pages and builds a long report in 2 to 4 minutes.
  • Labs creates full projects, like documents, presentations or small apps, from an idea. It requires a subscription.
  • Comet is its browser with the assistant built in, launched in July 2025 and free for everyone since October 2025.

How is it different from Google and ChatGPT?

All three use AI, but the biggest difference is the format of the result and how visible the sources are.

PerplexityGoogleChatGPT
What you getA written answer with numbered citationsA list of links (and sometimes an AI Overview on top)A written answer, with or without web search
Where the data comes fromVisible on top and in each claimThe links are the resultShown when it turns on search
How you talk to itNatural language and follow-upsKeywords, more and more full sentencesNatural language and follow-ups
Where it works bestResearch with citationsBrowsing, shopping, finding a siteWriting, creating, chatting

ChatGPT also searches the internet now, so the gap with ChatGPT has narrowed a lot. What still sets Perplexity apart is its focus on showing where each fact comes from, which is exactly what makes it useful for research.

Is Perplexity reliable?

Perplexity is as reliable as the pages it finds. If there's only one site with the fact for a query and that site is outdated, the answer will be wrong with a lot of confidence. That's why checking the citations is still your job.

It has also had its controversies. Dow Jones and the New York Post sued it, among other things, for attributing to them information they never published, and in 2025 security flaws were reported in its Android app. On privacy, you can turn off AI data retention in the settings so your queries aren't used to train models, and you can use incognito mode.

And on the website side, Cloudflare published in August 2025 that it detected undeclared crawling by Perplexity, with 3 to 6 million daily requests from a user agent that imitated Chrome, on top of the 20 to 25 million from its declared bot. Cloudflare removed it from its list of verified bots. Perplexity replied that it was a "sales pitch" and that the traffic wasn't theirs (Cloudflare's report).

How does Perplexity choose which sources to cite?

This is the part I find most useful if you work in marketing.

An Ahrefs study of 15,000 prompts found that 28.6% of the URLs Perplexity cites are in Google's top 10 for that same prompt, while for the rest of the assistants the overlap is around 8% (the Ahrefs study). The curious part is that it doesn't use Google's index, because its crawler, PerplexityBot, builds its own.

My take is that Perplexity rewards more or less the same things as Google (crawlable pages, relevant to what's being searched, from sites with authority) because it's built to cite. In other words, a page that already ranks is well on its way. Careful not to read it backwards, though: lists that look alike don't guarantee that ranking on Google gets you a citation.

Perplexity uses two bots, and it's worth telling them apart (official documentation on its crawlers).

  • PerplexityBot crawls pages for the index it shows in its results. The company says it doesn't use it to train models and publishes its IP ranges.
  • Perplexity-User visits a page when a user asks for something that needs that site, and since a person requested it, it generally ignores robots.txt rules.

How to show up in Perplexity if you have a brand

This blog's Search Console already shows searches like "tool to measure appearances in Perplexity", so the question is out there. This is what I'd do, in order.

  1. Check your robots.txt. If you block PerplexityBot, your pages don't get into its index and it's hard for it to cite you. That decision should be made with data, looking at your logs to see how often each bot comes by and which pages it visits (here's how to do a log file analysis and what to block in robots.txt).

  2. Work on the pages that already rank. Given the overlap with Google's top 10, good old SEO (crawling, indexing, specific content) is your best bet for Perplexity to cite you.

  3. Write passages that can be cited on their own. In a study I did of 5,000 AI answers, the cited pages had one heading every 100 to 200 words (the ones that barely showed up, one every 400 or more), 58% used question-style headings and 47% had an FAQ section. That study mixes assistants, so take it as a general trend and not as a rule for this search engine.

  4. Measure with a fixed panel of prompts. To start, 40 to 60 prompts of four types (the problem, your brand, comparisons and "alternatives to") are enough, run every month and several times each, because the answers change. Track presence, position, how it describes you and which sites it cites (share of voice in AI). The visits that do arrive show up in GA4 with perplexity.ai as the referrer (AI traffic in Google Analytics).

If you have to pick just one thing for this week, open Perplexity, ask it which company it recommends for what you sell and write down who shows up and which sources it uses. That list is your first work plan, and it costs nothing.

Free weekly resource

Get the enterprise SEO playbooks before anyone else

Every week we publish a new free resource: interactive audits, calculators, checklists and templates we actually use with enterprise clients. Leave your email and we'll send it to you first.

No spam. Unsubscribe whenever you want.

Enterprise SEO consulting

Want to apply this to your brand?

We design end-to-end SEO and GEO programs connected to pipeline and revenue. Senior team, weekly execution, executive reporting.