How can I check if AI chatbots recommend my products?
Ask ChatGPT, Claude and Perplexity the questions your buyers actually ask, without your brand name in them, several times each and from a neutral session. Then record whether you are named, in what position, and which brands and sources appear instead. Repeat the same questions every month to see the trend.
Why can't I just ask ChatGPT about my brand?
Because your customers don't. They ask category questions such as "which accounting software suits a five-person agency?", and the assistant decides which names to put in the answer. Asking "what do you think of my brand?" tests something else: whether the assistant knows you, not whether it recommends you.
There is a second trap: an absent brand is sometimes simply invisible to the assistant. OpenAI states that sites opted out of its search crawler, OAI-SearchBot, "will not be shown in ChatGPT search answers" [1], and Anthropic explains that blocking Claude-SearchBot keeps your content out of its search index [2]. Check access before you read anything into a result.
How do I check it, step by step?
Build a fixed list of questions, run it the same way on each assistant, and write everything down. The method matters more than the tool.
- Write 10 to 20 buying questions. Mix recommendations ("best…"), comparisons ("X or Y for…"), alternatives ("an alternative to…") and price questions. Never include your brand name.
- Check that assistants can read your site. Look for
OAI-SearchBot,ChatGPT-User,Claude-SearchBotandClaude-Userin your robots.txt and in your CDN or firewall rules [1] [2]. - Use a neutral session. A temporary chat or a logged-out window, so memory and custom instructions don't steer the answer.
- Ask each question three to five times per assistant. Answers change between runs; one answer is an anecdote, five give you a rate.
- Record five things per answer: whether you are named, your position in the list, which brands appear instead, which sources are linked, and the date.
- Repeat monthly with the same list. A stable list is what turns snapshots into a trend.
A spreadsheet with one row per answer is enough to start:
| Question | Assistant | Run | Cited? | Position | Cited instead | Sources shown |
|---|---|---|---|---|---|---|
| Best invoicing tool for freelancers in France? | ChatGPT | 1 | No | – | Brand A, Brand B | 2 comparison articles |
| Best invoicing tool for freelancers in France? | ChatGPT | 2 | Yes | 3 | Brand A, Brand B | 1 comparison article |
| Best invoicing tool for freelancers in France? | Claude | 1 | No | – | Brand A | None |
Illustrative rows. The brands are placeholders.
What should I measure?
Track how often you are cited, where, and against whom, then connect it to visits. Each metric answers a different question.
| Metric | How to compute it | What it tells you |
|---|---|---|
| Citation rate | Answers that name you ÷ answers collected | How often you are in the answer at all |
| Average position | Your rank among the brands named, when cited | Whether you are the first suggestion or a footnote |
| Share of voice | Your mentions ÷ mentions of all brands | Your weight against named competitors |
| Sources cited | Pages the assistant links or quotes | Which sites shape the answer, and where to be mentioned |
| Referred visits | Visits whose referrer is an assistant | Whether citations turn into traffic |
The last line matters because these visitors convert. In one client case study, Seer Interactive measured a 15.9% conversion rate for visits from ChatGPT, against 1.76% for Google organic, with 10.5% for Perplexity and 5% for Claude [3].
Which tools can automate the check?
Once the method works by hand, software can run it on a schedule. Three kinds of tools exist, and they suit different teams.
| Approach | Good for | Limits |
|---|---|---|
| Manual spreadsheet | Learning the method, a first baseline, no budget | Slow; easy to drift from a neutral setup |
| AI visibility dashboards such as Profound, Otterly.AI and Peec AI | Marketing teams that want scheduled tracking across many assistants in one dashboard [4] [5] [6] | They report the problem; changing your pages is a separate job |
| Ralator, from your coding agent (MCP) | Teams who also fix the site: measurement, audit, content plan and writing brief in Claude Code, Cursor or VS Code | Needs an MCP-capable client; full measurement is in the paid plans |
Profound describes itself as an AI marketing platform covering ChatGPT, Perplexity, Claude, Gemini, Copilot and Google AI Overviews [4]. Otterly.AI presents its product as a way to see "where your brand shows up on AI Search" [5]. Peec AI calls itself "AI search analytics for marketing teams" [6].
What if AI chatbots don't recommend me?
Then the measurement has done its job: it tells you which questions to work on. Fix access first, then the pages that answer those questions, then your presence on independent sites. In an Ahrefs study of 75,000 brands, mentions of a brand across the web showed the strongest correlation with visibility in Google's AI Overviews (0.664) [7].
Two guides go further: what generative engine optimization is and how to start, and how to improve your brand's visibility in AI search results.
Frequently asked questions
How many questions should I test?
Ten to twenty well-chosen buying questions are enough to see a pattern. What matters more is keeping the same set from one month to the next, so that changes reflect your work and not a new list.
Why do I get a different answer each time I ask?
Assistants generate each answer anew, and those using web search may read different pages each time. That is why one answer proves little: ask several times and measure how often you are cited.
Does my own ChatGPT account bias the result?
It can. Memory, custom instructions and past conversations personalise answers. Use a temporary chat or a logged-out session so the assistant answers as it would for a new customer.
Sources
- OpenAI. Overview of OpenAI crawlers.
- Anthropic. Does Anthropic crawl data from the web, and how can site owners block the crawler?
- Seer Interactive. Case Study: 6 Learnings, 1 site — How Traffic from ChatGPT Converts. 3 June 2025.
- Profound. tryprofound.com, product description.
- Otterly.AI. otterly.ai, product description.
- Peec AI. peec.ai, product description.
- Linehan, L. (Ahrefs). An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied). 26 May 2025.