Revenue AI comes wrapped in impressive numbers. Three that you will hear often:
- Gartner (2024): sellers who effectively partner with AI tools are 3.7 times more likely to meet quota than those who don't (survey of 1,026 B2B sellers).
- LinkedIn (2025): sales professionals using AI daily are 2x as likely to exceed their targets.
- Salesforce (2026): 87% of sales organizations use some form of AI.
They are real, sourced findings. But read them carefully.
Association is not cause
The first two are surveys that compare groups. Sellers who use AI well may also be the more organized, more curious, better-managed sellers. The survey cannot separate "AI made them better" from "better sellers use AI". Note the word effectively in the Gartner finding — it is the quality of use, not the licence, that is associated with results.
Predictions are not measurements
Gartner also predicts that by 2028 AI agents will outnumber sellers by 10 times, yet fewer than 40% of sellers will say AI agents improved their productivity. That is a forecast, and a warning: more AI does not automatically mean more productivity.
Vendor results are a starting point
Vendors publish their own customers' results. They are useful signals but they are chosen, not random, and the baseline is often unclear.
What counts as proof in your company
The only evidence that should change your plan is your own comparison:
- Record the baseline (e.g. meetings booked per rep, win rate).
- Run AI with one group and keep a comparable group without it (or with the old process).
- Agree the success threshold before you start.
- Compare after enough time for deals to move — weeks for meetings, quarters for win rates.
Treat survey numbers as reasons to test, not reasons to buy at scale.