WhatTheBotSaid logoWhatTheBotSaid

Guides · Updated October 21, 2026

How to Check if ChatGPT Recommends Your Brand

See where you stand first. The free check asks ChatGPT, Perplexity, Gemini and Claude five buyer questions about your category and shows who they name and which pages they cite. No account needed.

Run the free check

Most founders check by typing their category into ChatGPT once, seeing their name (or not), and moving on. That single answer is a weak signal. It comes from your own account, with your own history, on one day, from one assistant.

The short version: ask buyer questions without your brand name, in clean temporary chats, with web search on. Record who gets named and which pages get cited. Run each question more than once, check the other assistants too, and repeat on a schedule.

Step 1: Write the questions your buyers actually ask

Five to ten questions is enough to start. Good ones share three traits:

Mix intents: a “best X for Y” question, an “alternative to [big competitor]” question, a problem-first question (“how do I…”) and a price question.

Step 2: Use a clean chat with web search on

Your normal ChatGPT account is a biased test environment. Two settings matter:

If your buyers are in one country or city, put the location in the question rather than relying on where you happen to be.

Step 3: Record more than “yes or no”

For every answer, write down:

  1. Were you named? And in what position: first pick, in the list, or a footnote.
  2. Who was named instead. These are the brands the assistant sees as your competition, which may not match your own list.
  3. Which pages were cited. Open every source. The pages that name competitors but not you are your best targets, because the assistant already trusts them.
  4. Whether the description is right. Being named with the wrong price or the wrong audience can cost you as much as not being named.

A simple spreadsheet works: one row per question, per assistant, per run.

Step 4: Ask each question more than once

The same question can get a different list on a different day. Run each question at least two or three times, on different days, before deciding anything. Count how often you’re named across runs (“2 of 3”) rather than reacting to a single reply.

Step 5: Check the other assistants separately

Perplexity searches the web for its answers and shows its sources (Perplexity Help). Gemini grounds some answers in Google Search, and Google says not every response includes sources (Gemini Apps Help). Claude can search the web when web search is turned on (Claude Help). Each one works from different sources, so each gives its own answer.

From our own checks: we asked 5 questions a buyer of a free online form builder might ask to four assistants (20 answers, October 20, 2026). How often each one named Tally:

  • Gemini: 5 of 5 (answered without web search)
  • Perplexity: 3 of 5 (searched the web)
  • Claude: 2 of 5 (searched the web)
  • ChatGPT: 1 of 5 (searched the web)

Same brand, same questions. Google Forms was recommended in 13 of the 20 answers. On one question, Perplexity cited Tally’s own page comparing free form builders and listed Tally first, while ChatGPT named five other tools. Our checker calls Gemini through its API without Google Search grounding, so the Gemini app, which can search, may answer differently.

Step 6: Repeat on a schedule and track the trend

Keep the same questions and re-run them weekly or monthly. Track:

When the numbers are low, the guides on getting recommended by ChatGPT, Perplexity, Gemini and Claude cover what to fix for each one.

If you’d rather not do this by hand, WhatTheBotSaid’s free check writes five buyer questions for your category, asks ChatGPT, Perplexity, Gemini and Claude, and shows who each one named, which pages they cited, and whether you made the list. Paid plans re-run the same questions every week so you can see the trend.

Common mistakes

FAQ

How do I check if ChatGPT recommends my brand?

Write five to ten questions a buyer would ask without knowing your name, such as “best [category] for [type of customer]”. Ask each one in a new temporary chat with personalization off, make sure the answer used web search, and record who ChatGPT named, in what order, and which pages it cited. Repeat the same questions every week or month.

Why does ChatGPT give me a different answer than my customer?

Answers vary between runs, and they can be personalized. OpenAI says ChatGPT may use saved memories and your approximate location when it rewrites a question into web searches, and that temporary chats can be set to Unpersonalized so they don’t use memory or custom instructions. Your own chats also tend to be full of your brand, which skews what you see.

Should I put my brand name in the question?

Not for this test. “Is [Brand] any good?” tells you what ChatGPT knows about you, which is useful, but it doesn’t tell you whether ChatGPT recommends you to someone who hasn’t heard of you. Keep brand names out of the buyer questions and check reputation separately.

How many times should I ask each question?

At least two or three runs per question per assistant, on different days, before drawing conclusions. One answer is a snapshot. A pattern across runs, such as being named in 2 of 3 runs, is far more reliable than any single reply.

Do I need to check Perplexity, Gemini and Claude too?

Yes, if your buyers use them. The assistants often disagree. In one of our checks, the same form builder was named in 5 of 5 Gemini answers but only 1 of 5 ChatGPT answers to the same questions. Being named by one assistant tells you little about the others.

Is there a free tool that checks this for me?

Yes. WhatTheBotSaid’s free check asks ChatGPT, Perplexity, Gemini and Claude five buyer questions about your category through their APIs and shows who each one named, which pages they cited, and whether you made the list. No account is needed.

Sources