How do I check if ChatGPT recommends my products?

Why product-level checking is different

Most AI-visibility advice stops at the brand: does the assistant name your store? For an ecommerce business that's not enough. Buyers don't ask “is [your store] good” — they ask “best [product type] for [use case] under [budget],” and the assistant answers with specific products. Your brand can be well-known while a specific SKU never surfaces, or a single hero product can carry all your AI visibility while the rest of the catalogue is invisible. So the useful question is per-product: for the queries that matter, does this item show up?

The manual method, step by step

• Choose your priority SKUs — the products where an AI recommendation would actually move revenue.
• Write buyer-style shopping prompts — “best [category] for [need],” “[category] under [price],” “alternatives to [popular product].” Avoid naming your own product; you want to see if the model reaches for it unprompted.
• Run each prompt in every engine — ChatGPT, Perplexity and Gemini answer shopping questions differently, and some surface shopping results directly.
• Record three things: whether your product appears, how it's described, and which retailer or source is cited as the basis.
• Re-run several times. AI answers are volatile — 40–90% of the sources an engine cites can change when the same question is re-asked — so one result proves little.

This gives you a real snapshot, but by hand it's slow and easy to skew with a lucky prompt, and it tells you nothing about the trend or whether a mention actually earned a sale.

Turning spot checks into tracking

The scalable version runs those prompts continuously and reports your per-SKU visibility over time. This is where steek is purpose-built for stores: it tracks per-SKU AI Shopping visibility across ChatGPT, Perplexity and Gemini and ties it back to real revenue in GA4, so you see not just whether a product is surfaced but whether that surfacing made money. And because AI answers are noisy, every metric carries a 95% confidence interval — so a jump in a product's visibility is a signal you can act on, not sampling noise you chase.

That revenue tie-in matters because AI-referred traffic is famously hard to see in analytics — a large share arrives with no UTM, as “dark” traffic — so a tool that connects the AI mention to the product and the sale closes a gap most dashboards leave open.

What to do with what you find

If a priority product isn't surfacing, the fixes rhyme with GEO generally: make sure AI crawlers can read the product pages, give each SKU clear, accurate, self-contained detail (specs, use cases, honest comparisons), keep product facts consistent across your site and the retailers that list you, and earn mentions on the review sources models trust for that category. Then re-measure. For the ecommerce-specific playbook, see AI visibility for ecommerce, and for the engine-by-engine tactics, how to rank on ChatGPT.

To sample where you stand in a couple of minutes, run the free checker on the steek.ai homepage — no signup — and see whether the assistants your customers use put your products in the answer.