ChatGPT, Gemini and Google AI Mode don't recommend the same brands

By Carson Cobb. Data collected November 2026.

Most brands that check their AI visibility check one tool, usually ChatGPT, and assume the others behave the same way. So I asked ChatGPT, Gemini and Google AI Mode the same 48 buyer questions, in fresh sessions with no history, and compared every brand each one named.

They agree less than most people assume.

The short version

• Only 23% of the brands named came from all three AIs. 53% were named by just one of them.
• They picked the same top brand 15 times out of 45. Any two of them agreed on the first pick about 4 to 5 times in 10.
• Gemini and AI Mode name more brands than ChatGPT. Median per answer: Gemini 6, AI Mode 5, ChatGPT 4.
• ChatGPT plays it safest. Only 21% of its brands were named by neither of the others, against 34% for Gemini and 36% for AI Mode.
• Repeat runs of ChatGPT agree with each other more than ChatGPT agrees with Gemini or AI Mode. The gap between tools is bigger than the randomness inside one tool.
• Growing Shopify brands held up well across tools. When one of them was named, it was named by all three 41% of the time, against 18% for every other brand.

How much the three AIs overlap

45 of the 48 questions got brand recommendations from all three AIs. The other 3, about ingrown hair, perimenopause and a drink for focus, got advice and no brands from any of them.

Across those 45 questions, the three AIs produced 419 brand mentions once each brand is counted once per question.

Put another way: if you look at the full list of brands any AI recommended for a product, roughly half of that list depends on which AI the shopper happened to open.

There was almost always a shared core. 43 of the 45 questions had at least one brand that all three AIs named. Only 2 had none in common.

Pair by pair

For each pair of AIs, I measured what share of all the brands they named for a question were named by both.

ChatGPT and Gemini were the closest pair. Google’s two products, Gemini and AI Mode, were no closer to each other than either was to ChatGPT, which surprises people who assume they share one brain. AI Mode leans heavily on what ranks in Google’s results and on the product cards it shows, while Gemini writes longer, more opinionated lists.

Top picks travel better than the rest of the list

The first brand an AI names is the one shoppers are most likely to act on, so I checked how often each AI’s top pick showed up anywhere in the other two answers.

Being first on one AI is a good sign you’re at least in the answer on the others. It’s not a sign you’re first there too. All three put the same brand first in fewer than a third of questions.

How many brands each AI names

ChatGPT gave the shortest lists and the most mainstream ones. Gemini went wider, often adding a craft or specialist tier. AI Mode was the most likely to surface a brand neither of the others mentioned, often a product that ranks well in Google Shopping or a brand site Google’s results lean on.

For a brand, this means each AI has a different shape of opportunity. ChatGPT has fewer seats, and they go mostly to brands the editorial lists already agree on. Gemini and AI Mode have more seats and more room for a brand that isn’t the obvious pick.

Which product types agree the most

The share of brands named by all three AIs ranged from 16% to 33% depending on the product type.

Supplements and home goods had the most agreement, likely because a small number of testing sites and well known review roundups dominate those products. Food and pet had the least. Coffee subscriptions, tea, dark chocolate and dog treats produced long lists with very different picks on each AI.

Growing Shopify brands were more consistent, not less

I flagged brands that run their own store on Shopify and appear in my dataset of growing US brands.

Gemini and AI Mode put a growing Shopify brand first more often than ChatGPT did.

The more interesting number is consistency. When a brand from my dataset was named for a question, all three AIs named it 41% of the time (35 of 85). For every other brand it was 18% (61 of 334). The brands that earn their place, through reviews, testing sites and a clear product story, tend to earn it everywhere. The questions were written around product types these stores sell, which makes them more likely to appear than in a random set of questions, so treat this as a pattern worth checking for your own products, not a law.

Why the answers differ

Each AI builds its answer from different material.

• ChatGPT answers from its own knowledge plus a web search that leans on review sites, testing sites and magazines. See the sources ChatGPT cites.
• Google AI Mode pulls from Google’s own results, Reddit and YouTube, product listings and brand websites. In my answers it linked to brand websites more than any other kind of source. See where AI sends shoppers.
• Gemini writes longer answers that often mix a well known tier with a craft or specialist tier, and shows fewer links.

Different inputs produce different shortlists. A brand that is well covered by editorial roundups but weak in Google’s results can do well on ChatGPT and disappear on AI Mode, and the reverse.

What to do with this

1. Check all three, not one. The tool your customer opens decides half of who they see. A ChatGPT only check misses most of the picture.
2. Use your customers’ real questions. Run the same questions on each AI and record who’s named and in what order.
3. Look at the brands named by all three. That shared core is the bar for your product. Find what they have in common: the review sites, testing roundups and Reddit threads behind them.
4. Treat each AI as its own channel. AI Mode rewards what ranks in Google and clear product data on your own site. ChatGPT rewards editorial coverage. Gemini sits in between.

If you want to see where your products stand on all three, that’s what my AI Search and Conversion Audit covers. Or start with the free AI Visibility Check.

How this was done

• Questions: 48 buyer questions, 8 each for apparel, beauty and skincare, food and beverage, home, pet and supplements, written in everyday language. The same questions as Who ChatGPT recommends. They were written around product types sold by stores in my 94 store sample.
• How they were asked: ChatGPT on a paid plan with the default model, in a new temporary chat for each question. Gemini on the default model in a new temporary chat for each question. Google AI Mode in a new thread for each question, signed in, with no apps connected. ChatGPT answers were collected on November 4, 2026, and Gemini and AI Mode answers on November 5, 2026.
• Brands: every brand named in the answer as an option to buy, matched against a list of known brands and aliases for each question and checked by hand. Retailers, publishers and sources shown as citations were not counted as recommendations. Brands named only in Google’s link previews below the answer were left out. Brands named only as part of the question were left out. Because this counts every brand named as an option, including ones mentioned in passing, ChatGPT’s counts here are slightly higher than in Who ChatGPT recommends, which counted the ranked list.
• Overlap: for each pair of AIs and each question, the number of brands both named divided by the number of brands either named, averaged across questions.
• “Brands from my dataset”: brands with a store among the US Shopify stores behind my other reports.
• Limits: one run per AI per question, from one location and one set of accounts. AI answers change between runs, which is covered in how often AI recommendations change.

No store or brand is named in this report.