Market-share trackers rank AI engines by how many people use them. They say nothing about whether an engine gives the same answer twice. In Zumi's run of 8,740 answers across 304 beauty-category questions, a brand mentioned in one answer was mentioned again on the next run 54.7% of the time on ChatGPT, 58.1% on Perplexity, and 62.3% on Google AI Overviews. The gap between engines is real, and so is the gap between any engine and a stable result.
Which AI engine repeats a brand mention most often?
Zumi collected answers for 304 questions in the India beauty and personal care category over 11 days (2026-07-23 to 2026-08-02). Repeat runs exist for three engines. The measure is simple: when a brand was mentioned in an answer, was it mentioned again in the next run of the same question on the same engine? The median gap between runs was 24 hours.
| Engine | Mentioned again on the next run | Cases |
|---|---|---|
| ChatGPT | 54.7% | 17,873 |
| Perplexity | 58.1% | 9,715 |
| Google AI Overviews | 62.3% | 12,306 |
| All three | 57.9% | 39,894 |
Google AI Overviews held a mention 7.6 points more often than ChatGPT. That is a useful difference for deciding how many repeat runs to buy, because the least stable engine needs the most observations before one result can be trusted. It does not make any engine safe to read from a single check.
Limits of the measure: one category, one market, 11 days, and three engines. Gemini, Claude, and Copilot had no repeat runs in this data, so nothing here ranks them. The figures describe what the engines mentioned. They say nothing about traffic or revenue.
Which AI engine has the largest market share?
It depends on what is counted. Statcounter counts referrals that AI chatbots send to websites. First Page Sage estimates monthly active users in the US. Both are recent, and they disagree most on Gemini.
| Engine | Statcounter (September 2026, worldwide referrals) | First Page Sage (October 2026, US users) |
|---|---|---|
| ChatGPT | 80.48% | 52.1% |
| Gemini | 10.91% | 26.9% |
| Claude | 0.95% | 10.6% |
| Perplexity | 5.85% | 1.9% |
| Microsoft Copilot | 1.79% | 1.3% |
Gemini and Claude take far more of the user count than of the referral count. That is the useful signal: an engine can have an audience without sending that audience to websites, so referral analytics alone undercount where buyers are asking. Both trackers move monthly, so the figures above are dated snapshots.
Which AI engine drives the highest-converting traffic?
Studies disagree on size. Seer Interactive found ChatGPT referrals converting at 15.9% against 1.76% for Google Organic, and Microsoft Clarity, measuring 1,200+ publisher sites, found about 11x. The edge over organic search depends on the site, the industry, and how conversion is counted. AI referral traffic remained a small share of total traffic through 2025, so the multiplier matters less than the volume it applies to.
Does visibility on one engine carry over to another?
No. A September 2026 academic audit of 15 commercial prompts across ChatGPT, Copilot, Google, and Perplexity found that 96.4% of cited URLs appeared in only one engine. Combined with the repeat-mention figures above, two things hold at once: a result on one engine does not predict another, and a result on one day does not predict the next.
How should marketers prioritize AI engines?
Three inputs decide the order, and a market-share table supplies only the first.
- Where the buyers are. Referral analytics show which engines already send visitors. User-count trackers show which engines have an audience that does not click through.
- How stable the result is. The less often an engine repeats a mention, the more repeat runs are needed before a change in the result means anything.
- Whether the engines agree. Because few citations overlap, each engine needs its own reading. Gemini, AI Overviews, and AI Mode cite differently for the same buyer question.
Zumi is an AI Search Intelligence Platform built to measure that variance directly: it tracks mention rate, share of voice, average position, and citation share across up to nine AI engines, scoped to the ones a given brand's buyers actually use. Book a demo to see the breakdown for a specific category.
Sources
- Zumi. Brand-mention repeat rates, India beauty and personal care prompt-run corpus (8,740 answers, 304 questions, 2026-07-23 to 2026-08-02). Method and tables in what an AEO checker can and cannot show.
- Statcounter. "AI Chatbot Market Share Worldwide" (September 2026). gs.statcounter.com
- First Page Sage. "Top Generative AI Chatbots by Market Share" (October 2026). firstpagesage.com
- Seer Interactive. "How Traffic from ChatGPT Converts" (2025). seerinteractive.com
- Microsoft Clarity, via Digiday. "The state of AI referral traffic in 2025." digiday.com
- Tannenbaum, B. "Scoring With the Engine: Retrieval Exposure, Cross-Engine Divergence, and the Limits of Engine-Agnostic GEO Scores" (arXiv, September 19, 2026). arxiv.org