Home / Growth Circuit AI / Measured pilot
Choose a single playbook, such as win-back, and one channel. Name the one metric that decides success, for example a purchase within 14 days.
Agree the audience, the time window, the control group size and the smallest lift that would count as a success. Writing it down first keeps everyone honest.
A random slice, 10% by default, gets no message. Because it is random, it shows what would have happened anyway.
Send with approvals on and a frequency cap. Do not change the audience or message halfway, or the comparison stops being fair.
Put the counts into the calculator below. You get the lift and a range, not just a single number. Then scale it, change it or drop it.
| Item | Decision |
|---|---|
| Playbook and channel | One of each, for example win-back by push |
| Primary metric | One metric, for example a purchase within 14 days |
| Window | The full conversion window plus a short buffer |
| Control group | A random 10% by default, bigger for low conversion rates |
| Success threshold | Agreed before the send |
| Approver | A named person signs off every send |
| Sharing results | Only with your written consent, names removed |
The starting numbers are made up so you can see how it works. Replace them with your own.
Ten percent is a common default. When conversion rates are low, a bigger control group gives a clearer answer, because a small one makes the result noisy.
Long enough for the conversion window to finish, plus a short buffer. For a purchase within 14 days, wait until the 14 days are over for everyone before you compare.
If the control group is random and the rules were set beforehand, yes. The calculator shows a range as well as a single figure, so you can see how sure the result is.
You do. We only share a result publicly with your written consent, with names removed.
Tell us the platform and the playbook you would start with. We reply within two working days.
Request a pilot