How to Track Your Products in Google Shopping and ChatGPT Carousels
To track your products in Google Shopping and ChatGPT carousels, use one fixed set of shopping questions for both and run each several times. Record how often each product appears, whether it is near the front, who the seller is, and whether the price and details are right. Tracking the two together makes sense because ChatGPT's carousels draw heavily on Google Shopping. In a Semrush experiment of 100 runs, ChatGPT's top carousel product was among Google Shopping's first three results 75% of the time. Another analysis reported that 83% of carousel products came from Google Shopping. ChatGPT now also accepts product feeds directly from merchants. Because carousels change from run to run, positions are misleading. Presence rates across repeated runs are the reliable measure.
Where do ChatGPT carousel products come from?
When a question calls for products, ChatGPT runs searches in the background to build its answer. Semrush found that for answers with product carousels it runs two sets of searches. One set informs the written answer. The other is a set of shopping queries sent to Google Shopping, whose results largely shape which products appear in the carousel. A separate analysis reported by Search Engine Land found that 83% of carousel products came from Google Shopping data.
OpenAI also invites merchants to share product feeds for ChatGPT shopping, which gives it product data directly from retailers. So in practice, carousel inclusion depends on:
- your products' presence and strength in Google Shopping, through Merchant Center feeds and product pages;
- your ChatGPT merchant feed, if you participate;
- your own product pages, which ChatGPT also crawls: in Profound's study of 812,190 product cards from June 2026, 87.3% drew on web crawling and 12.7% on merchant feeds;
- the product details, reviews and prices those sources contain.
How ChatGPT ranks shopping results is covered in how ChatGPT ranks shopping results.
Why do ChatGPT Shopping carousels change so much that rankings are hard to track?
- Different shopping searches each run. ChatGPT rewrites the question into shopping queries, and the wording varies, which changes the Google Shopping results it sees.
- Live data. Prices, stock and offers change daily, and products drop out when unavailable.
- Personal context. Location, earlier conversation and memory can change the question and the products shown.
- Generated answers. The final selection and order are produced fresh each time.
- Feed updates from merchants add and remove listings.
The result is that the same question can show a product first in one run and not at all in the next. A single observation tells you little. Measuring how often a product appears across many runs is stable enough to track over time.
What should you measure instead of rankings?
| Metric | Definition | Why it matters |
|---|---|---|
| Presence rate | Runs where the product appears ÷ total runs | The core visibility measure |
| Front-row rate | Runs where it appears in the first three positions ÷ total runs | Early positions get most attention |
| Your-store rate | Appearances where your store is the seller ÷ all appearances of the product | Whether you or a reseller wins the listing |
| Accuracy rate | Appearances with correct price, name and key attributes ÷ all appearances | Wrong details lose sales |
| Competitor share | Competitor product appearances ÷ all product appearances for the question | Who you lose to |
Report these per question and per product group, with the number of runs.
How do you track products in Google Shopping?
- Use Merchant Center performance reports for impressions and clicks on your products across Google's shopping experiences, and diagnostics for disapprovals. In the US, Canada, Australia, India and New Zealand, its AI performance insights also show how visible your brand is in AI Mode, AI Overviews and the Gemini app against similar brands.
- Track your shopping questions in Google's Shopping results with a rank tracker or SERP data service that supports Shopping and location settings.
- Record the top results for each question: products, sellers and prices.
- Check Google's AI Mode for the same questions, since it also shows products from Google's shopping data.
If your products are missing from the first page or two of Google Shopping results for the shopping queries ChatGPT runs, they are unlikely to appear in its carousel: in the Peec AI data reported by Search Engine Land, about 60% of strongly matched carousel products came from Google Shopping's top 10.
How do you track products in ChatGPT carousels?
- Write 20–50 shopping questions in buyers' words, with the constraints they use: price, size, features, use.
- Run each question three to five times, in fresh chats, from the country you sell in.
- Record every carousel item: product name, seller, price, position and any rating shown.
- See the shopping searches ChatGPT ran. Semrush describes a method using the browser's developer tools. Open the Network tab, reload the conversation and search the responses for "search_model_queries" to see the queries it generated.
- Run those queries in Google Shopping and compare the results with the carousel.
- Repeat monthly and after feed or price changes.
Tools that track ChatGPT shopping results can automate the runs once your question set is stable. Diagnosing gaps by hand is covered in how to diagnose product feed gaps in ChatGPT shopping with manual searches.
How do you diagnose why a product is missing?
| What you find | Likely cause | What to do |
|---|---|---|
| Not in Google Shopping's top results for ChatGPT's shopping queries | Weak feed data, missing attributes, category, price or reviews | Improve titles, attributes, identifiers and reviews in Merchant Center |
| In Google Shopping, not in ChatGPT's carousel | Constraints in the question not matched, or other products favoured | Check that attributes answer the question's constraints; compare prices and ratings |
| Shown, but via a reseller | Reseller listings stronger or better matched | Check identifiers, pricing and feed quality for your own listing |
| Shown with wrong price or details | Stale feed or mismatched product page | Automate price and stock updates; align page and feed |
| Appears rarely and randomly | Borderline relevance | Strengthen the attributes and content for that use case |
How to structure the underlying product data is covered in how ecommerce brands can structure product data for AI recommendations.
What does a tracking round look like in practice?
Consider an illustrative outdoor gear retailer that tracks 30 questions such as "best waterproof hiking boots for wide feet under £150". It runs each four times in ChatGPT and checks Google Shopping for the shopping queries ChatGPT generated. Its best-selling boot appears in 38 of 120 carousel runs, a presence rate of about 32%, and in the first three positions in 14 of them. For the "wide feet" questions, though, it appears in only 2 of 24 runs.
The shopping queries explain why. ChatGPT searched for "wide fit waterproof hiking boots", and in Google Shopping the retailer's listing had no width attribute and a title that did not mention wide fit. The retailer's product page did mention it, but only in a paragraph of prose. After the retailer adds the width attribute to the feed and includes "wide fit" in the title for the relevant variants, the product appears in Google Shopping's top results for those queries. The next month it appears in 11 of 24 runs. The carousel was simply reflecting the feed.
What do people ask about tracking shopping carousels?
How many products does a ChatGPT carousel show?
Typically several, often around eight, though the number varies by question.
Does being in ChatGPT's merchant programme replace Google Shopping?
Not on current evidence. Google Shopping data still shapes much of what carousels show, so keep Merchant Center strong as well.
Should I track carousel positions at all?
Track the front-row rate rather than exact positions; exact order changes too much between runs.
How often should shopping visibility be checked?
Monthly for trends, and after major feed, price or catalogue changes.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources