ChatGPT and Google AI recommend different brands 62% of the time, BrightEdge study finds

Three platforms, three different answers

A BrightEdge study ran tens of thousands of identical prompts through ChatGPT, Google AI Overviews and Google AI Mode using the firm's AI Catalyst tool. According to a dev.to analysis of the results, the three systems recommended different brands on 61.9% of queries, and only 17% of prompts produced the same brand recommendations across all three. Just 33.5% of queries prompted a brand mention from every platform at all.

The platforms also diverge on how often they name any brand. ChatGPT returned no brand on 43.4% of queries, while Google AI Overviews mentioned at least one on all but 9.1% of them.

Mentions and citations point in opposite directions

The study's most counterintuitive finding concerns the gap between naming a brand and linking to it:

Platform	Brand mentions per query	Citations per query
ChatGPT	2.37	0.73
Google AI Overviews	6.02	14.30
Google AI Mode	1.59	9.49

ChatGPT names roughly 3.2 brands for every citation it produces, meaning it will recommend a product in its answer without linking to that company's site. Google's systems invert that pattern. AI Overviews attaches about 2.4 citations for every brand it names, and AI Mode is the most link-heavy of the three, generating roughly six citations per mention.

The dev.to write-up reads this split as structural rather than incidental: ChatGPT behaves like a recommender leaning on knowledge absorbed during training, whereas Google's answers operate as retrieval systems that link to fetched pages, which are not necessarily the brands the system would endorse. That implies two distinct optimization targets, the write-up argues — being present in the model's knowledge for ChatGPT, and being present in Google's retrieved source pool.

Agreement depends on the question

The disagreement is not evenly spread across query types. BrightEdge measured same-brand agreement by intent and found a wide range: "compare" queries aligned 80% of the time, "buy" queries 62%, "where" queries 38% and "best" queries only 23%. In other words, when users ask for the "best" product in a category, the three platforms name different brands more than three quarters of the time. The write-up credits query structure for the spread: a comparison has a small, bounded answer set, while an open-ended "best" question leaves each platform's ranking model to fill in the blank differently.

Healthcare sees the most divergence

Industry shapes the numbers as well. Disagreement was highest in healthcare at 68.5%, followed by education at 62.1% and B2B tech at 61.7%. Finance, at 57.9%, and e-commerce, at 57.1%, were the most aligned sectors, yet still diverged on well over half of queries.

Why it matters

The findings undercut the growing habit of treating "AI search" as a single channel that can be optimized once and scored with one visibility metric. If the three major AI answer surfaces disagree on brand recommendations 62% of the time, strong visibility in ChatGPT says almost nothing about visibility in Google's AI results, and vice versa.

The practical takeaway for marketing and SEO teams is to audit each platform separately: ask ChatGPT directly whether it mentions your brand in category queries, then inspect which sources Google's AI Overviews actually link to. Because of the mention–citation gap, these are different tests with different outcomes. The write-up also flags a more basic distinction: a site that AI crawlers can reach is not the same thing as a site those systems recommend, and the 62% disagreement rate suggests recommendation is won or lost platform by platform.