How to Check if ChatGPT Recommends Your Company (Step-by-Step)

More buyers ask ChatGPT for a recommendation than click through a list of search results, and the answer usually names two or three companies. If yours is not one of them, the lost deal never shows up in your analytics. This guide shows how to check, by hand in about 30 minutes, and how to keep checking every day without the manual work.

Short answer: ask ChatGPT (and Gemini, Grok) 10 questions a new buyer would ask, without your brand name, in a fresh signed-out chat, with your market in the question, and record whether you are named, your position, which rivals are named and which sources are cited. Repeat on three days. To do this daily, use an AI monitor. If ChatGPT names someone else, read why ChatGPT recommends your competitor.

Step 1: Write buyer questions, not brand questions

Asking "What is Qomvia?" tests knowledge. Asking "What is an affordable tool to monitor my website's readiness for AI agents?" tests recommendation. Mix five types:

Category

"Best [category] for [audience] in [country]"

Alternatives

"Alternatives to [market leader]"

Problem

"How can I [job to be done]?"

Comparison

"[You] vs [rival]"

Local / price

"Affordable [category] under [price]"

Step 2: Remove personalisation

Use a private window, signed out where possible, a new chat per question. Memory and history bias answers toward brands you've asked about.

Step 3: Ask every model you care about

ChatGPT, Gemini and Grok answer differently. In one of Qomvia's own runs for "How can I track which competitors AI assistants recommend instead of my company?", Claude named Qomvia in one sample and a rival in another, same model, same day.

Step 4: Score each answer

Step 5: Repeat and calculate a mention rate

Mention rate = answers naming you ÷ answers sampled. Qomvia's own rate was 3.3% in one run: we track ourselves with the same method. See how to measure AI search visibility.

Step 6: Diagnose why you are missing

1. Can the model read you? Run the free agent-readiness score. Check client-side rendering and bot protection.
2. Are you in the sources it cites? If a comparison page is cited and you're absent, that's the gap.
3. Is your entity clear? See entity SEO.
4. Can you be cited? See how to get cited by ChatGPT.

Example: a worked check for a B2B SaaS

Say you sell AI visibility software in Switzerland. Ten questions might be:

1. Best tools to track brand visibility in AI assistants
2. Affordable alternatives to Profound and Peec AI
3. How can I check if my website is ready for AI agents?
4. Which competitors do AI assistants recommend instead of my company?
5. Cheapest AI visibility tracker for a small business
6. AI brand monitoring tool that tracks Grok
7. Qomvia vs Semrush AI toolkit
8. How do I monitor my products in ChatGPT shopping?
9. GEO tool for agencies with several clients
10. Tool to see which sources ChatGPT cites in my category

Run each in ChatGPT, Gemini and Grok on three days = 90 answers. If you're named in 9, your mention rate is 10%.

How to read the results

Named but not cited

The model knows you but uses other sources. Work on pages worth citing, see how to get cited by ChatGPT.

Cited but not named

Your page is a source, but the answer recommends others. Make your offer and entity explicit (Organization JSON-LD and sameAs).

Neither

Check readability first: why blocking AI crawlers costs you answers, training, search and user-fetch bots.

Named with wrong facts

Fix the source pages and publish clear, dated facts (dates that agents can read).

ChatGPT vs Gemini vs Grok: why answers differ

Each assistant decides when to search the web, which index it uses and how it formats lists. ChatGPT may search the web for time-sensitive questions; Gemini answers in the Gemini app and separately in Google's AI Overviews and AI Mode; Grok draws on the web and X. Compare like for like: same question, same day, fresh chat. Our AI search engines field guide covers each.

Manual vs automated checking

Start manual to learn the questions; automate once you know which ten matter.

Prompt templates you can copy

Replace the brackets and keep the wording fixed across rounds:

• "What are the best [category] for [audience] in [country]?"
• "Which [category] would you recommend for a [company size] with a budget under [price]?"
• "What are alternatives to [market leader] for [use case]?"
• "Is [your brand] a good choice for [use case]? What are the alternatives?"
• "Which [category] do experts recommend for [problem]?"
• "Where can I buy [product] in [city] with delivery tomorrow?" (for shops)

Questions with a constraint (budget, country, size) are closer to purchase and more useful than generic "best X" prompts.

Signals that predict whether ChatGPT will name you

Across the sites in Qomvia's index, the businesses that get named tend to share a few traits we measure in the 26-check score:

1. Server-rendered content that agents can read without JavaScript.
2. Crawler access for search and user-fetch bots in robots.txt (content signals).
3. Clear entity data: Organization JSON-LD, consistent name and address.
4. Machine-readable summaries: an llms.txt file, sitemaps and feeds (sitemaps for agents).
5. Presence in third-party sources, comparisons, directories and reviews the models cite.

For our own site, the score is 99/100, yet the mention rate was only 3.3% in the same run. Readability is necessary, not sufficient: point 5 is where most brands, including us, still have work to do.

How often to re-check

Weekly is enough for a manual panel; daily is better for decisions because single days swing. After you ship a fix, wait at least two weeks before concluding anything, and compare the same questions on the same models.

Step 7: Automate it

Manual checks cost about 30 minutes per model per round. Qomvia's AI monitor asks 11 tracked questions to ChatGPT, Gemini and Grok every day and shows the rival that won each answer, from $69/month per site.

Turn the check into a daily habit with Qomvia

A manual panel tells you where you stand today. Qomvia tells you every day and shows what to change. Start with the free agent-readiness score: it checks whether ChatGPT and other agents can read your site at all, and lists a fix for every failing check. The free Site monitor keeps re-scanning weekly.

Then add the AI monitor: your buyer questions go to ChatGPT, Gemini and Grok daily, with mention, position, sentiment, sources and the rival that won each answer. Sign up free or see plans and pricing.

Questions

Can I ask ChatGPT directly if it recommends my company?

You can, but it tests knowledge, not recommendation. Ask buyer questions without your brand name.

Why does ChatGPT give different answers each time?

Answers are sampled and may use live search, so lists vary. Repeat each question and use a mention rate.

How many questions should I test?

Start with 10 across category, alternatives, problem, comparison and price questions.

Does being mentioned mean ChatGPT recommends me?

Not always. Read the sentence around your name; a neutral or negative mention still counts as named.

Is there a free way to check if AI mentions my brand?

Yes: the manual method above is free, and Qomvia's agent-readiness score is free.

How do I get ChatGPT to recommend my business?

Make your site readable to agents, be present in the sources it cites, keep entity facts consistent and track progress over time.