A Perplexity visibility tracker asks Perplexity the questions your buyers ask and records whether each answer names your brand. Depra asks Sonar, the answer model Perplexity sells to developers, with live web search set to your country, once a week, and stores every answer with your position, the tone of the mention and the websites it cited.
Depra reads Perplexity through its API. It does not capture the Perplexity app. In Depra's India study, Perplexity was the engine whose brand list changed most when the same question was asked in Hinglish instead of English. Depra runs the same questions on ChatGPT, Gemini and Google AI Overviews.
How does Depra capture Perplexity answers?
Depra sends each tracked question to Sonar through the Perplexity API, with web search on and the search biased toward the country you track. Sonar searches the web, writes the answer and returns the list of websites it used. Depra stores the answer and every source.
This is an API capture. Depra does not record the Perplexity app, where model choice and settings can differ, so an answer you see in the app may not match the one Depra stores.
Checks run once a week on every plan, in any of 40 markets and up to 10 per project. Location is set at country level: Depra does not target cities or add personas.
Why weekly?
Cost. Perplexity charges a flat fee on every search request. A Perplexity check costs Depra more than a ChatGPT or Gemini check, so running it daily would raise the price of every plan. ChatGPT, Gemini and Google AI Overviews stay daily.
Weekly checks mean fewer answers behind each Perplexity number: about 87 a month for 20 questions in one market on Starter, against about 600 for each daily engine. Your report shows a confidence range next to each engine, so a smaller sample shows up as a wider range.
Perplexity was also the most consistent engine between reruns in Depra's study: two runs of the same question in the same language shared 68.3% of their brands, against 54.7% on ChatGPT and 51.2% on Gemini.
| Engine | How often | Checks a month |
|---|---|---|
| Perplexity | Weekly | About 87 |
| ChatGPT | Daily | About 600 |
| Gemini | Daily | About 600 |
| Google AI Overviews | Daily | About 600 |
What you see for Perplexity
For each weekly Perplexity answer, Depra records the numbers below for you and each competitor you track, from 3 to 20 per project depending on plan. In Depra's India study, Perplexity gave the longest source lists of the three engines studied: 19.6 sites per English answer on average, against 9.4 for Gemini and 2.5 for ChatGPT. Those gaps are one reason a brand can appear on one engine and not another, covered in why AI engines disagree about your brand.
- Visibility: the share of weekly Perplexity answers that name your brand, across the date range you choose.
- Position: your spot in Perplexity's recommendation list. Position 1 means Perplexity put you first.
- Sentiment: positive, neutral or negative, quoted from the Perplexity sentence that mentions you.
- Cited websites and pages: every site labelled by type (editorial, forum, marketplace, directory and more), and every cited page labelled as a listicle, comparison, how-to, product page or other format.
- Change over time: each number compared with the previous period, last month or last year.
- Every answer stored: Perplexity's full reply, the brands it named and its numbered source list, with CSV export.
Sites that recommend your competitors on Perplexity
Depra's citation gap analysis lists the websites AI answers cite, including Perplexity's, when competitors are named and you are not, with competitors' own sites removed. Scoring rules: how Depra measures AI visibility.
Why does Perplexity change its brand list by language?
Depra's study shows how much it changes and does not prove why. Hinglish questions mix Hindi and English words typed in English letters, the way most Indians ask ('sabse accha protein powder kaunsa hai'). Across 480 answers collected from inside India on 14 August 2026, the study counted the brands two answers to the same question have in common, as a percentage of all brands either answer named.
Two Perplexity runs in the same language shared 68.3% of their brands. An English and a Hinglish answer to the same question shared 44.9%. That 23.4 percentage point drop is the largest of the three engines studied: 7.8 points on Gemini and 2.2 on ChatGPT.
Perplexity cited sources in every answer in both languages, so the change is in which sites and brands it picks. Cited websites moved even more: two runs shared 85.2% of their sites, and an English and a Hinglish answer shared 52.8%. In skincare, The Derma Co appeared in 2.5% of English answers and 20.0% of Hinglish ones, while La Roche-Posay fell from 13.8% to 1.3%.
For a brand selling in India, an English-only Perplexity score does not show what Hinglish buyers get. Depra suggests Hinglish versions of your questions when India is one of your markets, and your report splits visibility by English and Hinglish. Method and full tables: English vs Hinglish AI shopping study.
Brands shared by two Perplexity answers to the same question: 68.3% in the same language, 44.9% across English and Hinglish. (Depra Hinglish study, 480 answers, India, 14 Aug 2026)
Can PerplexityBot read your site?
PerplexityBot is the crawler Perplexity uses to show and link websites in its answers, and Perplexity's bot documentation says it is not used to train AI models. Perplexity recommends allowing it in robots.txt, the file that tells bots which pages they may visit, if you want to appear in its results.
A second agent, Perplexity-User, fetches a page when a person asks about it, and Perplexity says it generally ignores robots.txt rules. Depra's AI crawler checker shows the rule that applies to both and marks Perplexity-User as advisory, because a block rule may not stop it. The guide to AI crawlers covers the other AI bots Depra checks.
How to check Perplexity by hand
You can sample Perplexity yourself before you buy a tool. The spreadsheet method for tracking brand mentions in AI answers has the columns and scoring rules.
- Pick 10 buying questions. Leave brand names out. If you sell in India, add a Hinglish version of each.
- Ask in a new thread. Use perplexity.ai and note which model is selected, since answers can vary by model.
- Log the answer and its sources. Record named or not, position, tone and the numbered sources.
- Repeat weekly. Compare the share of answers that name you from week to week.
Where the manual method stops working
The app answer depends on the model and settings in your account, and logging every numbered source by hand takes time. Depra records the Sonar answer from inside your chosen country each week and keeps every source.