Ask ChatGPT Twice for a Brand's Top 3 Alternatives and the List Changes 83% of the Time

A marketing manager asks ChatGPT for the best alternatives to her company and takes a screenshot of the three competitors it names first. A colleague asks the same question that afternoon and gets a different three. Both screenshots end up in the same strategy deck, and they disagree.

That matters because "alternatives to [brand]" is how buyers build a shortlist, and the first three names get the most attention. For a brand, the competitors an AI engine names next to it are the competition it faces in AI search. If that list moves every time someone asks, a single check tells you very little.

It moves most of the time. Asked twice, ChatGPT names a different top three alternatives for 82.9% of brands, and Gemini for 93.6%. Where a market has clear leaders, such as auto marketplaces, the list holds; almost everywhere else it shifts.

On September 23, 2026 we asked ChatGPT and Gemini "What are the best alternatives to [brand]?" for 500 brands, the brands whose websites appear most often in Google results for 20,000 common searches. Each question was asked twice on each engine, a few hours apart. We compared the brands where both answers named at least three alternatives: 414 on ChatGPT and 456 on Gemini.

The top three are the first three alternatives an answer names. A brand's top three counts as the same when both answers name the same three competitors first. We also counted how many of the first answer's top three came back in the second answer's top three, whether the first-named alternative stayed the same, and whether the first answer's top three appeared anywhere in the second answer.

ChatGPT named the same top three in both answers for 17.1% of brands. On average it kept 1.26 of its three leading alternatives from one answer to the next, and it opened with the same first alternative 38.2% of the time. For most brands, the competitor ChatGPT names first depends on when you ask.

Part of the change is order and part is new names. 55.3% of the competitors in ChatGPT's first top three were still named somewhere in its second answer, in a different slot or further down. The other 44.7% did not appear at all, replaced by different companies. The top of an AI list is where a buyer's attention goes, and it shifts from one ask to the next.

Gemini named the same top three for 6.4% of brands, and on average repeated 0.96 of its three leading alternatives, fewer than one. A brand checking its competitors on Gemini sees an almost new shortlist every time.

A brand's competitive set in AI search is therefore a range of possible lists. The competitors that matter are the ones that keep coming back across many asks, not the three in any single answer.

Where a category has clear leaders, ChatGPT repeats itself. For Best Buy it named Amazon, Walmart and Costco both times. For Autotrader it named CarGurus, Cars.com and CarMax. For Bankrate it named NerdWallet, WalletHub and LendingTree, and for Avalara, Vertex, Sovos and TaxJar.

By industry, auto dealer brands had the most stable lists, with the same top three for 47.6% of brands, nearly three times the overall rate, followed by accounting (35.3%) and banking and loans (28.2%). These are markets led by a few well-known national names. In crowded, fragmented markets, the top three is open.

Do not judge your AI competitive set from one screenshot. Ask the same alternatives question several times, on more than one engine, and count how often each competitor appears. A competitor that shows up in most of your checks is a real rival in AI search, even if it was missing the one time you looked.

If you are the challenger, the instability is an opening. In most markets the top three is not locked, so coverage on the comparison and review sites AI engines read can move you into the list. In stable markets such as auto marketplaces and big-box retail, breaking in takes a sustained presence in the sources every answer draws on.

Track it over time rather than once. Working with an ai seo company that runs repeated checks across engines shows which competitors AI names next to you and how often.

Embed this research

Cite this study