84% of AI Brand Recommendations Change Daily, Study Finds

A Supercite pilot study running the same 210 buying questions against ChatGPT, Gemini, Google AI Overviews, and Google AI Mode on consecutive days found that 84% of answers named a different set of brands the next day. The research, officially released on November 18, 2026, reveals that 27% of brands recommended on day one disappeared entirely in day-two answers, with cited source pages rotating even faster than the brand lists themselves.
The finding challenges the assumption that AI search engines deliver consistent product recommendations. According to the study, checking whether an AI engine recommends your brand once provides minimal signal, because asking the same question 24 hours later frequently produces the opposite result.
How Brand Lists Rotate Overnight
Supercite documented a representative example from Google AI Mode. When asked for the "easiest CRM for a team that has never used one," the engine recommended Less Annoying CRM first on day one, citing its "zero-frills simplicity" based on a Zapier list of best small business CRMs. On day two, Less Annoying CRM vanished from the answer entirely, replaced by HubSpot as the top pick. This time the engine cited two different pages: a guide on Slack's blog and HubSpot's own product page.
Nothing about the CRMs changed overnight. What changed was which pages the engine happened to read. The study found that an AI answer is not a fixed ranking but a summary of a few pages picked fresh each time. Swap the pages and you swap the answer.
Volatility by Engine and Category
Gemini changed the most, with 95% of questions producing a different brand list the next day and dropping a third of previously named brands. Google AI Overviews changed the least, though still two out of three answers moved between days. ChatGPT showed the highest top-pick loyalty, naming the same number-one brand 71% of the time, likely because it often starts with a shortlist from memory before reading external pages.
Category stability varied widely. Accounting software barely moved, with QuickBooks, Xero, Wave, and FreshBooks appearing consistently across engines. Running shoes had the most unstable top pick, changing on two out of three questions. The least stable category was AI search visibility tools, a new category with many young companies and no settled set of authoritative lists.
Source Page Rotation Outpaces Brand Rotation
Only 42% of pages cited on day one were cited again on day two, while brands showed 61% overlap. This gap exists because the same market leaders appear on many different lists, so even when the cited pages rotate, the leading brands remain visible. Smaller brands face a different reality: brands ranked lower in an answer retained their mentions only 51% of the time, compared to 73% retention for brands that appear in many answers across their category.
The pattern is structural. A leader appears on most of the pages an AI reads, so whichever pages it picks, the leader is on them. A smaller brand appears on a few. When the engine picks other pages, the smaller brand drops out. The study found that visibility for non-leaders depends entirely on which pages the engine reads that day.
What This Means for Brand Visibility
A single screenshot of ChatGPT recommending or ignoring your brand is one sample from a distribution, not a verdict. Across eight answers to the same question (four engines, two days), the study found an average of 14 different brands mentioned, with 30% of them named only once. Only about two brands per question appeared in all eight answers.
The study recommends measuring share of mentions across many answers over time, not yes-or-no presence in a single check. For brands that are not category leaders, the fix is to get onto more of the "best X" lists that AI engines cite. The more of them that name you, the less it matters which ones the engine reads on any given day. The study also recommends checking cited sources, not just the final answer, because source pages rotate even faster than brand lists. Supercite's platform automates this daily tracking across all four engines, delivering first results the same day users start monitoring.
on Emergent today
.avif)




