FFeedlegible
Guides

How to check whether AI assistants recommend your store — and why one check isn't enough

Asking ChatGPT "what's the best X?" once tells you less than you think. We re-asked the same shopping questions two weeks apart. Whether a store was named barely moved; who else got named changed a lot. Here's a simple way to track it properly.

Varun Mehta · November 12, 2026 · 3 min read

Most store owners check their AI visibility the same way: open ChatGPT, type "best [what I sell]", and see if their name comes up. It's a good instinct. But one answer is a snapshot, and the snapshot moves.

What changed when we asked again

We re-ran the same shopping questions for three stores, about two weeks apart, across ChatGPT, Gemini and Perplexity — 18 repeat answers in all. A small sample, but the pattern was clear:

  • Whether the store was named barely changed. In 17 of 18 repeat answers, the store was either named both times or missed both times. The one change: Perplexity named an organic bedding brand for "best sheet sets" on 13 October, and didn't on 22 October.
  • Who else was named changed a lot. On average, only about a third of the brands in an answer were the same two weeks later. In 12 of the 18 repeat answers, fewer than half the brands carried over. Three Perplexity answers shared no brands at all with the first time.

So the assistants are fairly consistent about whether you're in their picture — and much less consistent about the rest of the list. A single check can make a competitor look dominant, or make you look closer to the list than you are.

A simple way to track it properly

1. Write down 5–10 real questions. Use your customers' words, not your category name. Mix broad ("best duvet covers") with specific ("best linen duvet cover for hot sleepers under $200"). The specific ones are where a smaller brand can actually win.

2. Ask each one in every assistant your customers use. At minimum ChatGPT, Gemini and Perplexity. They don't agree with each other: in our audits a store named by one assistant is often missing from the other two.

3. Use a clean session. Log out, or use a private window. Assistants that remember you, or know where you are, give different answers to different people.

4. Record three things for each answer:

  • Were you named?
  • Were you linked (a clickable source or product card)?
  • Were you the top pick, or one of several?

Add the other brands named, and the sources cited if the assistant shows them. Perplexity shows sources by default; in ChatGPT and Gemini you can ask "what sources did you use?".

5. Repeat monthly, and read the trend, not the day. Your named / not-named count across 30 answers means something. One answer on one day doesn't. If a brand shows up in most of your answers month after month, that's a real competitor for AI attention. If it appears once, it was probably noise.

What moves the numbers

From the stores we've audited, two things matter most:

How we measured

  • Three Shopify stores (home, bedding and kids' goods) checked between 13 October and 6 November 2026; each repeat was 5–20 days after the first.
  • Same question, same assistant, asked as a shopper in the store's market. We compared which brands were recommended by name.
  • 18 repeat answers is a small sample: treat the exact percentages as indicative.

Want the first round done for you? The free audit asks the three assistants your store's two most likely questions and shows who's named instead. Feedlegible's Growth plan re-asks your own questions every week.

Is ChatGPT recommending your store?

Free audit in about a minute: what AI assistants recommend for what you sell, and what AI agents can't read on your store. No login.

More from the blog