What is AI Win-Rate?
AI Win-Rate™ is how often AI systems select a brand head to head at the decision stage.
It is measured as the share of buying conversations, run through to a decision, in which an AI assistant or agent recommends the brand outright. Shared picks, no-picks and referrals elsewhere do not count as wins.
Also written AI win rate or AI win-rate. AI Win-Rate is AIVO's metric, defined by AIVO, Inc. and published as AI Win-Rate: A Board Metric for AI-Mediated Buying (Methodology v2.0).
A buying conversation, run to a decision
- The unit is a conversation, not a prompt. Each run is a full, multi-turn exchange in which an AI assistant or agent researches and compares options for a written buyer brief, then is asked to choose one.
- Four assistants. ChatGPT, Gemini, Claude and Perplexity, with the model version and web-search state logged on every turn.
- Repeated runs, with confidence ranges. AI answers vary between runs, so each scenario is run several times and results carry 95% confidence ranges.
- Five outcomes, one counted. Every conversation ends as won, shared, lost to a competitor, lost to an intermediary such as a retailer or adviser, or no pick. Only outright wins count toward AI Win-Rate.
- The same scenarios each reading. Readings re-run an unchanged scenario set, so a change in the number reflects a change in the AI systems, not in the questions.
Full rules: Methodology v2.0 and the method page.
How AI Win-Rate differs from other AI visibility metrics
Most AI visibility metrics count whether a brand appears. AI Win-Rate counts whether it is chosen.
| Metric | What it counts | When in the journey | What it does not show |
|---|---|---|---|
| Mention rate or appearance rate | How often a brand appears anywhere in AI answers | Any answer | Whether the brand is chosen |
| Citation rate | How often a brand's pages are cited as sources | Sources behind an answer | Whether the cited brand wins the recommendation |
| Share of AI voice | A brand's share of mentions against competitors | Across many answers | Share of decisions |
| PSOS (Prompt-Space Occupancy Score) | Presence and favorability in first answers to a standardized prompt set | First prompt | What happens after follow-up questions |
| AI Win-Rate | Outright selection when the buyer asks the AI to choose | The decision | On its own, the effect on sales, which the AI Win-Rate Study is testing |
The two measures often disagree. In AIVO's Beyond Visibility working paper, 87.3% of brands named at the first turn of a buying conversation had been displaced by a competitor by the fourth (1,427 brand probes, ten industries, 95% CI 85.5 to 88.9%). The cause AIVO identifies is the Linkage Gap: the model holds facts that would support the brand but does not carry them to the decision.
Does AIVO weight partial wins?
No. AI Win-Rate is deliberately strict: only an outright recommendation counts. A shared pick, a hedged answer, or a referral to a retailer or adviser is reported separately, not scored as a fraction of a win. That keeps the headline number simple and comparable across brands, categories and readings.
The detail sits alongside it. Each run records the turn at which the brand was displaced (the Displacement Initiation Turn), how it held up when the buyer challenged the recommendation, and the sources the AI relied on. For a graded score, ARQ (AI Recommendation Quotient) combines recommendation at the decision (50%), resilience under challenge (25%) and head-to-head performance (25%) on a 0 to 100 scale, and is used for AIVO's public AI Rankings.
Where AI Win-Rate comes from
AIVO developed the metric in its published research during 2026. Each step below links to a dated, DOI-registered working paper on Zenodo.
- 19 February 2026. AI Recommendation Intelligence (ARI) sets out outcome-based measurement of AI-mediated decisions, with the final-recommendation win rate as a core metric alongside conversational survival.
- 21 February 2026. Conversational Survival Rate shows that a brand can be mentioned early in an AI conversation and eliminated before the final recommendation.
- 14 April 2026. The AIVO Paradox reports that high AI visibility does not predict, and often coexists with, a low AI purchase-recommendation win rate, across 137+ brands in five categories.
- 13 May 2026. Agentic Brand Control names the discipline of measuring and managing whether an AI agent selects a brand and completes the task with it.
- 8 September 2026. AI Win-Rate: A Board Metric for AI-Mediated Buying (Methodology v2.0) publishes the full measurement rules for AI Win-Rate, superseding version 1.0 (DOI 10.5281/zenodo.22280231).
- 11 September 2026. The AIVO Standard Methodology, version 4.1 lists AI Win-Rate among its core decision-stage metrics.
- 23 September 2026. LLM Equity Valuation (L.E.V.), version 2.0 uses AI Win-Rate as an input to valuing a brand's AI-layer position.
- 30 September 2026. AIVO Win Rate launches as a self-serve product, letting any brand or agency measure its AI Win-Rate.
- 24 October 2026. AIVO Glossary, version 1.0 fixes one spelling and one definition for AI Win-Rate.
Sources
- AI Win-Rate: A Board Metric for AI-Mediated Buying (Methodology v2.0), AIVO, Zenodo.
- Beyond Visibility, the Linkage Gap working paper, AIVO, Zenodo.
- How AIVO measures AI Win-Rate, the method page.
- AIVO Glossary, with every AIVO term and retired names.
- Client case: a skincare brand's AI Win-Rate from 38 to 58 in five months.
What is your AI Win-Rate?
Measure it, fix the top three causes and re-measure in 30 days with the AI Win-Rate Sprint, US$3,500, or talk to us about enterprise measurement.
See the Sprint Talk to us