Four AI apps, four different shelves
AEOsome sent the same shopping questions to ChatGPT, Google AI Mode, Copilot and Perplexity. Product cards showed up on very different shares of answers, and the products barely matched.
AEOsome Research published its AI Shopping Visibility Study 2026 on Oct 11, 2026. It is long, with ten findings. This post covers the one that matters most for a brand: the shelf is different in every app.
How the study was run
The study has two parts, 30,855 answers in all. Study 2 is the one used here. In May 2026, AEOsome sent 5,584 shopping questions across seven categories (electronics, beauty, health, home, sports, toys and auto) to ChatGPT, Google AI Mode, Microsoft Copilot and Perplexity. After removing repeat attempts, that left 22,335 answers.
A "product card" is a structured product in the answer with a name, a price and a merchant link. A brand named in a sentence does not count.
How often each app shows product cards
Google AI Mode showed a product card on 64.4% of queries. Copilot did on 44.5%, ChatGPT on 28.5% and Perplexity on 5.0%. AEOsome says the app matters more than the question: the spread across apps is 59 points, and across the nine shopping intents it is 40.
The ChatGPT figure is a mixture. AEOsome says 60% of ChatGPT’s answers came from a mini model that showed a card 3.7% of the time. Its full models showed one 65.3% and 72.2% of the time. The model was not controlled in the study.
The same question, different shelves
Of 5,583 questions asked on all four apps, 1.6% got a product card on all four, and 21.9% got one on none. The most common result, 31.7%, was a card on exactly one app.
Where two apps both showed cards, the sets of products overlapped very little. AEOsome reports a Jaccard index of 0.01 to 0.05 for products and 0.25 to 0.42 for brands. In plain words, the apps often named similar brands but pointed to different products.
Each app reads a different web
AEOsome also typed the sources each app cites. For Google AI Mode, 35.1% were community content, such as YouTube and Reddit. For Copilot, 36.5% were retail and 15.0% marketplace. For ChatGPT, 49.9% were editorial, and 14.5% of its answers cited no source.
Re-asking the same question also changes things. Card versus no card agreed across retries 89.8% of the time on Google AI Mode, 86.4% on Perplexity, 75.4% on Copilot and 57.6% on ChatGPT.
Limits to keep in mind
- All answers are US, English, logged out, captured from each app’s consumer web interface by an automated service.
- Personalization, location and mobile surfaces were not observed.
- ChatGPT’s model was not pinned, so its numbers mix several models.
- Brand and source typing is rule-based, with some brands and about 5% of domains left unresolved.
What to do this week
You cannot check one app and call it done. Tick these off as you go.
How Shopthru.ai helps
Shopthru.ai is an AI agents team for consumer brands. It asks ChatGPT, Gemini, Google AI, Perplexity and Copilot what your shoppers ask, finds where AI gets your products, prices or policies wrong, and drafts each fix into an approval inbox. It ships only on your yes, then checks again. Instagram and Amazon are next.