When AI recommends your brand, does it translate into sales?
AIVO is running a research study with a leading marketing scientist to test whether AI recommendation predicts commercial performance. We are inviting a limited cohort of established brands, balanced across categories, to take part, contribute outcome data and see the findings first.
Cited is not chosen. Does chosen mean sold?
Across 20,000+ multi-turn probes and 200+ brands, AIVO measured 87.3% displacement between being cited by an AI assistant and being chosen by it (working paper WP-2026-14). Brands can raise how often AI chooses them. Nobody has yet shown whether that produces more sales. The study applies the tests used for any marketing metric, starting with two questions.
Research, not a vendor claim
Executives are shown new marketing metrics constantly, usually by the companies selling the tools. This study tests ours with an independent academic, and the findings will be published whatever they show, including if Win Rate turns out to be a mirror of market share.
A Distinguished Professor of Marketing at a leading US business school, editor-in-chief of a leading international marketing research journal, a former president of the American Marketing Association's Academic Council, and ranked among the top 2% of scientists worldwide in Stanford University's database. His work links marketing metrics to sales, profit and shareholder value. Academic collaborators are paid for research time, not results, and payments are disclosed in the research.
Four reasons to take part
- Take part in the researchHelp investigate whether AI recommendation has predictive value beyond traditional brand and market metrics.
- Evidence your board can useSee how your AI Win Rate relates to your own sales, bookings or conversions, using controlled comparisons rather than visibility reports.
- Early accessConfidential cohort benchmarking and a private briefing on the findings before publication.
- Value regardless of the resultThe full 90-day measurement and improvement program, plus re-measurements at around 6 and 12 months.
How the 90 days run
One brand, one category, one market, measured across ChatGPT, Gemini, Claude and Perplexity. The unit of measurement is the probe: one multi-turn buying conversation, with the model version and web-search state logged on every turn.
- Weeks 1–2BaselineWin Rate against named competitors, with a 95% confidence interval. The held-back control is agreed and frozen.
- Weeks 3–11RemediationWe find where AI stops recommending you and why, then fix the evidence behind it. You approve every change.
- Weeks 12–13Re-measurementThe baseline probes are re-run under the same conditions, with the held-back and market controls.
- After day 90Outcome reportWin Rate before and after, what moved it, and what to keep doing.
Two controls separate the effect of the work from market movement: part of your prompt set or product range is held back until after day 90, and an untouched market prompt set nets out category-wide shifts, including model updates. Lift counts only above the upper bound of the baseline's confidence interval.
What cohort members share
- One monthly outcome figure for the measured category and market, such as sales, bookings, leads or account openings, for the 12 months before and after the start. No personal or customer-level data.
- A held-back control left untouched until after day 90.
- Consent to anonymized research use. Your brand is never named without your written agreement.
- A named owner and approvals within five business days, under a short data-sharing agreement signed before the baseline.
Study pricing, in context
Program fees pay for measurement and remediation. The research depends on the outcome data cohort members share, which is why cohort pricing is half the planned post-publication price.
| Study cohort | After publication | |
|---|---|---|
| Fixed fee | $9,500 | $19,000 |
| Success fee | $750 per point of qualifying lift | $1,500 per point |
| Success fee cap | $15,000 | $30,000 |
| Maximum total | $24,500 | $49,000 |
| Re-measurements at 6 and 12 months | Included | Quoted separately |
| Cohort benchmark and briefing | Included | Not available |
Your fee buys the program, not a result: findings are reported whatever they show, including where there is no lift. All measurement is external, with no access to your systems and nothing to install.
Discuss cohort eligibility
Places are limited so the study stays balanced across categories. A 30-minute conversation confirms fit and your outcome measure; the baseline starts within two weeks of agreement.
Get in touch Download the prospectus (PDF)