How to Check Whether ChatGPT and Perplexity Recommend Your Product Without Paid Tools
Amit Malvi
Founder, SearchAxe. Former Senior SEO Manager, Invideo AI
Published Updated 9 min read
Quick answer
You can check whether ChatGPT and Perplexity recommend your product in about an hour with no paid tools. Write ten prompts the way your buyers ask, run each in a logged-out or non-personalized session on both engines, repeat your top three prompts, and log who is named, who is recommended and which pages are cited in a spreadsheet. The patterns tell you whether to monitor, fix sources or correct misdescriptions.
On this page
- What do you need before you start?
- How do you check whether ChatGPT and Perplexity recommend your product?
- Which prompts should you use for a one-hour check?
- How do you read the results of a manual check?
- What do free reports add to a manual check?
- What are the limits of a one-hour check?
- What should you do after the check?
Key takeaways
- Ten well-written buyer prompts, run logged out on ChatGPT and Perplexity, are enough for a first read on whether you get recommended.
- Repeat your three most important prompts, because AI engines rarely return the same brand list twice.
- Log who is named, who is recommended for the stated need and which pages are cited, including absences.
- ChatGPT's citations overlap with Google's top 10 far less than Perplexity's do, which changes what you fix for each engine.
- Free first-party reports from Google and Microsoft, plus GA4, fill in what a manual check cannot see.
You can check whether ChatGPT and Perplexity recommend your product with a browser, a spreadsheet and an hour. If you're a founder who has just been told to "go ask ChatGPT about your brand", this is how to do it so the answer means something: prompts that match how buyers ask, sessions that are not personalized to you, a few repeated runs and a simple way to read the results. It covers two engines so it fits in an hour; the full set to watch is ChatGPT, Perplexity, Claude, Gemini, Google AI Overviews, Google AI Mode and Copilot.
What do you need before you start?
You need five things, none of them paid: a blank spreadsheet, a private browser window, a list of four to six competitors that show up in your deals, a one-paragraph fact sheet of what must be true about your product, and an uninterrupted hour. Neither engine needs a paid plan. OpenAI says ChatGPT web search works for people who are not signed in, and Perplexity's help center describes a free tier with a daily allowance of Pro Searches.
OpenAI's help page on searching the web with ChatGPT confirms the logged-out option and adds that ChatGPT may search automatically when a question would benefit from current information. The same Perplexity help page says it searches the web in real time and every answer includes numbered citations to its sources. Being logged out matters: OpenAI says memory and approximate location can shape the searches ChatGPT runs, and your own account has probably heard a lot about your product.
How do you check whether ChatGPT and Perplexity recommend your product?
The one-hour manual check
- Write ten buyer prompts (8 minutes)Use the templates in the next section and the exact words buyers use on sales calls, so the test matches real demand.
- Set up the sheet (4 minutes)Columns for engine, prompt, run, named, recommended, competitors named, cited URLs and description errors, so nothing is judged from memory.
- Run all ten in ChatGPT (12 minutes)Use a logged-out window or an unpersonalized Temporary Chat, open a new chat for every prompt, and note whether the answer shows sources.
- Run all ten in Perplexity (12 minutes)Use a logged-out window and the default search, and copy the numbered citations next to each answer.
- Repeat your top three prompts twice more (12 minutes)Rerun the three shortlist prompts that matter most on both engines in fresh chats, because one answer is one draw.
- Score and read the patterns (12 minutes)Count mentions and recommendations per engine, list which competitors appear most, and highlight every misdescription.
The repeat step is not optional. In SparkToro and Gumshoe's research, as Search Engine Journal reported in January 2026, there was less than a 1% chance of ChatGPT returning the same list of brands twice for the same prompt. Three runs will not give you a precise rate, but they will show whether you are a fixture of the answer or an occasional guest.
Find out which AI engines recommend you
A free consultation on where you appear today across AI engines and search.
Schedule a Free ConsultationWhich prompts should you use for a one-hour check?
Use prompts a buyer types before they have a shortlist, plus a few that test how you are described. Weight the ten toward "best tool for my situation" prompts, because that is where recommendations are won or lost. Fill in your own category, buyer and rivals.
| Number | Intent | Template | Example |
|---|---|---|---|
| 1 | Shortlist | Best [category] for [buyer] | "Best AI customer support tool for a Shopify store" |
| 2 | Shortlist with constraint | Best [category] for [buyer] under [constraint] | "Best AI support agent for a store with two support staff" |
| 3 | Use case | How do I [job] with AI | "How do I automate order status replies with AI" |
| 4 | Use case | Tool that can [specific capability] | "AI tool that answers returns questions in WhatsApp" |
| 5 | Alternatives | Alternatives to [rival] | "Alternatives to Gorgias for small stores" |
| 6 | Alternatives | Cheaper or simpler than [rival] | "Simpler alternative to Zendesk AI for ecommerce" |
| 7 | Head-to-head | [You] vs [rival] | "Your product vs Gorgias" |
| 8 | Fit | Is [you] good for [buyer] | "Is your product good for a store doing 500 orders a month" |
| 9 | Brand | What is [you] | "What is your product and who is it for" |
| 10 | Pricing | How much does [you] cost | "How much does your product cost" |
How do you read the results of a manual check?
Read the sheet for four patterns, each with a different next step. Count first, then open the cited sources, because the sources explain the counts. The cited pages are the most useful thing a manual check produces.
| Pattern | What it usually means | Next step |
|---|---|---|
| You are not named on any shortlist prompt | Engines do not associate you with the category, or cannot find pages that say you belong in it | Check crawler access, then the pages and third-party lists that define your category |
| Named on brand prompts, missing on shortlists | Engines know you exist but not who you are best for | Publish clear use-case and comparison pages, and earn mentions on the roundups being cited |
| One rival appears in almost every answer | That rival is in the sources the engines retrieve most | List the cited URLs for their mentions and find the ones you are missing from |
| You are named but described wrongly | An outdated or inaccurate source is being retrieved | Find the cited page behind the error and update it, or publish a clearer page of your own |
Expect the two engines to disagree, and use that. Ahrefs found in its August 2025 study of 15,000 prompts that 8.0% of ChatGPT's in-text citations ranked in Google's top 10 for the same prompt, against 28.6% of URLs cited by Perplexity. If you appear in Perplexity but not ChatGPT, your pages that already rank well on Google and Bing are probably doing the work, since Perplexity's citations overlap most with top-ranking pages even though it runs its own index. ChatGPT's search, which OpenAI says sends rewritten queries to providers including Microsoft, leans more on Bing visibility and third-party mentions.
What do free reports add to a manual check?
Three free reports cover what manual prompts cannot. None of them shows whether ChatGPT or Perplexity named you, but together they show clicks from ChatGPT, impressions on Google's AI surfaces and citations on Microsoft's.
- GA4: ChatGPT adds utm_source=chatgpt.com to referral links, according to OpenAI's publishers FAQ, so those sessions and signups can be isolated. Our GA4 guide covers the other engines.
- Search Console: the generative AI performance report, rolled out to all sites on August 31, 2026, shows impressions of your pages in Google AI Overviews and Google AI Mode.
- Bing Webmaster Tools: the AI Performance report counts citations of your pages in Copilot and Bing's AI summaries and lists the grounding queries behind them.
Before reading anything into a zero, rule out a blocked crawler. OpenAI's crawler documentation says sites that block OAI-SearchBot will not be shown in ChatGPT search answers, and Perplexity's crawler page says PerplexityBot is the crawler that surfaces and links websites in its results. Check your robots.txt, CDN and firewall for rules like these, which allow both.
User-agent: OAI-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /What are the limits of a one-hour check?
It is a first read, not a measurement. Ten prompts and a few repeats is a small sample, logged-out answers can differ from what a signed-in buyer sees, and answers move week to week: SISTRIX's May 2026 study found ChatGPT replaced 74% of its cited sources each week. Two engines also leave five unchecked. Treat a strong result as a reason to monitor, not to stop.
What should you do after the check?
Decide based on the pattern you found. If you are recommended on most shortlist runs, turn the sheet into a monthly habit using how to track brand mentions across ChatGPT, Perplexity, Gemini and Google AI, and calculate share of voice in AI search against your rivals. If ChatGPT is the gap, the deeper method is in how to track brand mentions in ChatGPT, and why your brand does not show up in ChatGPT walks through the causes.
If the check raises more questions than it answers, such as why a rival wins everywhere or where a misdescription comes from, that is the job of a full AI visibility audit across all seven engines with every cause traced. If you would rather keep it in-house with software, compare the options in AI visibility tracking tools compared for startups.
Frequently asked questions
Can I check if ChatGPT recommends my product for free?
Yes. ChatGPT web search works for people who are not signed in, according to OpenAI, so you can run your buyers' prompts in a private browser window at no cost. Write about ten prompts, run each in a new chat, repeat the most important ones, and log who is named and which pages are cited in a spreadsheet. GA4 can then show clicks from ChatGPT through its utm_source=chatgpt.com tag.
Why should I log out before checking ChatGPT?
Because your own account is personalized. OpenAI says that when memory is on, ChatGPT may use saved memories when rewriting search queries, and it uses an approximate location from your IP address. If you have discussed your product with ChatGPT before, your results will not match a stranger's. A logged-out window or an unpersonalized Temporary Chat, which does not use memory or custom instructions, gives a fairer read.
How many prompts are enough for a manual AI visibility check?
Ten prompts, weighted toward shortlist and alternatives questions, are enough for a first read in an hour, as long as you repeat the most important three. They are not enough for a reliable rate: SparkToro and Gumshoe ran prompts 60 to 100 times per platform to measure how much answers vary. Use the manual check to find patterns, then move to monthly tracking with more runs if the stakes justify it.
Why does Perplexity recommend my product when ChatGPT does not?
The two engines retrieve sources differently. Ahrefs found that 8.0% of ChatGPT's in-text citations ranked in Google's top 10 for the same prompt, against 28.6% of URLs cited by Perplexity. Perplexity searches the web in real time on every answer and, although it runs its own index, tends to cite pages that already rank well on Google. ChatGPT may answer from training or search through providers including Microsoft, so Bing visibility and third-party mentions matter more.
