Frank Vitetta Research Finds AI Assistants Disagree on Brand Mentions in Buying Questions

London, UK, October 7th, 2026, FinanceWire

Study of 749 identical shopping questions, drawn from nearly 27,000 AI answers logged by LLM Scout, shows ChatGPT is the assistant most likely to name a brand when the others do not

London-based AI consultant Frank Vitetta has published new research showing that four of the most widely used AI assistants often return different brands in response to the same buying question. When ChatGPT, Claude, Gemini and Perplexity were given the same 749 shopping and buying questions, they disagreed on whether to name the tracked brand in 27.2% of cases, or more than one question in four.

The questions were drawn from almost 27,000 AI answers logged by LLM Scout over a nine-month period. The full research note, including the method and the limits of the study, is published on Frank Vitetta’s website.

AI assistants as a buying channel

A growing number of consumers and business buyers now use AI assistants to decide which product to buy or which company to use. In response, brands have started to invest in AI search marketing, a practice also referred to as generative engine optimisation (GEO) and AI optimisation (AIO), with the aim of being named in those answers.

The research indicates that a brand can appear in an answer from one assistant and be absent from another assistant’s answer to exactly the same question. As a result, the brands a shopper sees depend in part on which AI application they use.

Key findings

The study reported the following results across the 749 questions put to all four assistants:

• On 53.7% of the questions, none of the four assistants named the tracked brand.
• All four assistants named the brand on 19.1% of the questions, or about one question in five.
• Where at least one assistant named the brand, all four did so in only about four cases in ten.
• On 117 questions, only one of the four assistants named the brand. ChatGPT was that single assistant on 50 of those questions, about twice as often as any of the other three.
• Across all answers, ChatGPT named the tracked brand in about 36 of every 100 answers. Claude did so in about 28 of every 100, while Gemini and Perplexity each did so in about 24 of every 100.

Implications for businesses

According to the research, many companies assess their visibility in AI assistants by entering a small number of questions into ChatGPT alone. The findings suggest that this approach gives an incomplete view, because results from one assistant do not reliably predict results from the others.

“For a business, one assistant’s list is one opinion,” Frank Vitetta writes in the research note.

The research also points to an effect on buyers. Two colleagues researching suppliers with different AI applications may arrive at two different shortlists without being aware of it. “Neither list is wrong,” Vitetta writes, adding that each list represents only part of the picture.

Recommendations

Based on the findings, the research sets out several practical steps for brands and buyers:

• Brands should check how they appear across all four assistants, rather than relying on ChatGPT alone.
• Brands should record the questions where they are absent, as well as those where they are named.
• Brands should repeat these checks regularly, as all four assistants changed over the nine months covered by the study.
• Buyers who use AI to build a shortlist should put the same question to a second assistant and compare the results.

Notes to editors

• In this study, “named” means the answer mentioned the brand. It does not mean the assistant recommended it.
• Where no assistant named the tracked brand, the answers may still have named other companies.
• The study does not explain why ChatGPT names brands more often than the other assistants.
• Each assistant answered a slightly different mix of questions overall, so the four overall naming rates are not a strict like-for-like ranking. The 749-question comparison is like for like.
• The research covers a limited set of brands and buying questions. The data was collected with LLM Scout (https://llmscout.co). It is not public and has not been independently reviewed.

About Frank Vitetta

Frank Vitetta is an AI consultant based in London. He helps businesses choose AI tools and trains their teams to use them. His research examines how ChatGPT, Claude, Gemini and Perplexity name brands in their answers. Users can learn more information at frankvitetta.co

Website: https://frankvitetta.co/

Contact