How to check if ChatGPT recommends your products

Your customers are asking ChatGPT what to buy. Here is how to find out what it tells them.

The 5-minute manual check

You do not need any tools for this. Open ChatGPT and ask it the same questions your customers would ask.

Step 1: Write 5-10 buyer prompts

Think about how a real shopper phrases things. Not "Nike Air Max 270 review" — that is a Google search. An AI prompt sounds like:

• "What are the best running shoes for flat feet?"
• "I need a lightweight moisturizer for oily skin under $30"
• "Recommend a stroller that folds small enough for a city apartment"
• "What is the best alternative to [your competitor]?"
• "Compare [your product] vs [competitor product]"

Write them down. These are your test prompts.

Step 2: Run them across multiple assistants

ChatGPT is the biggest, but it is not the only one. Check at least:

Assistant	Why it matters
ChatGPT	Largest user base, strongest purchase intent
Gemini	Google's AI — deeply integrated into search results
Perplexity	Cites sources directly, high conversion traffic
Claude	Growing user base, detailed comparative answers
Copilot	Built into Windows and Edge, mainstream reach

Run the same prompts in each. Copy-paste the answers.

Step 3: Score each answer

For every response, note:

1. Were you mentioned? Yes or no. If no, that is a gap.
2. Which products were named? Specific SKUs or generic brand mention?
3. Were the details accurate? Price, features, availability.
4. Who else was mentioned? Which competitors showed up?
5. Was there a recommendation or just a list? Did the assistant actively recommend you or just include you in a lineup?

Step 4: Check the sources

In Perplexity and some other assistants, you can see which sources the AI pulled from. This tells you why you were or were not mentioned:

• If a competitor is always cited from a review site you are not on, that is a gap to fix.
• If your product page was cited but the details were wrong, your page needs better structured data.
• If no sources mention you at all, you need to build third-party presence.

What to do when you are not mentioned

Your product data is incomplete

AI assistants parse product pages for structured information: price, features, availability, dimensions, materials. If your page is vague or relies on images to convey specs, the assistant skips it.

Fix: Add complete, text-based specifications to every product page.

You lack third-party mentions

Assistants cross-reference multiple sources. If your brand only appears on your own website, the assistant treats it as a self-serving claim rather than a verified fact.

Fix: Get listed on review sites, comparison articles, and industry roundups.

Your FAQ does not match buyer language

If your FAQ says "product specifications" but shoppers ask "what is this made of?", the assistant will not connect the two.

Fix: Write FAQ answers in natural, conversational language. Match the phrasing real shoppers use.

Your competitors are optimizing for AI

Some brands are actively feeding clean data to AI assistants through structured markup, detailed product descriptions, and consistent information across the web. If they are doing it and you are not, they win by default.

Fix: Start a structured AEO program. It does not have to be complicated — even basic schema markup and clean product data goes a long way.

The problem with manual checks

Manual testing works for a quick audit. It does not work for ongoing monitoring because:

• AI answers change weekly as models update and the web changes
• Running 10 prompts across 5 assistants every week is tedious
• You miss changes between checks
• It is impossible to scale across hundreds of products

That is where automated monitoring comes in. SeenByLLM runs these checks daily across 8 assistants, tracks changes over time, and surfaces the gaps that matter most — so you can fix them before your competitors do.

Start tracking your AI visibility

Start a free trial and see what AI assistants are saying about your brand today. It takes about five minutes to get your first results.