Geo test readiness checker
Find out whether a geo holdout test can give you a clear answer about a channel, before you spend anything. Two numbers decide it. No account needed.
Will a test work for you?
Two numbers decide whether a geo test can give you a clear answer: your total order volume, and how much of it the channel you want to test claims.
Use the conversions your ad platform reports for that channel.
Based on simulations of UK regional sales data. Your exact threshold depends on how your sales spread across regions. The free designer checks that with your real data.
Why two numbers and not one
Most readiness advice asks for one number: your conversion volume. That misses half the question, and it lets brands run tests that were never going to work.
A geo test measures your total sales in regions where the channel is switched off, against regions where it kept running. So your total order volume sets the floor: the smallest change in total sales the test can pick out from normal week-to-week movement. More volume, smaller detectable change.
The channel you are testing sets the ceiling. If your ad platform credits a channel with 300 of your 20,000 weekly orders, then even if every single one of those is genuinely caused by the ads, the most the test could ever see is a 1.5% change in total orders. Your floor at that volume is about 3%, so the answer comes back inconclusive no matter how good the ads are. You would have paid for a test that could not have worked.
So the check is a comparison. The channel's share of your orders has to clear the detection floor your volume implies, with room to spare, because only part of what the platform credits is usually incremental.
The detection floor by volume
These are the bands the checker uses. They come from simulations on real UK regional sales data, scaled up and down with order counts redrawn at each volume.
| Total orders per week | Smallest change the test can detect |
|---|---|
| Under 200 | Not enough volume to test |
| 200 to 400 | About 18% of total orders |
| 400 to 800 | About 12% of total orders |
| 800 to 2,000 | About 7% of total orders |
| 2,000 to 8,000 | About 4% of total orders |
| 8,000 and above | About 3% of total orders |
Simulated sales are cleaner than real ones, so treat these as the best case. Your own figure also depends on how your sales spread across regions.
What else affects the answer
- How your sales spread across the country. A brand doing most of its volume in London is harder to measure than one spread evenly, because the rest of the country has to stand in for London and it behaves differently. No single region should be much more than a third of your orders.
- How much history you have. The test needs a stable run of past weeks to learn how your regions normally track each other. Around twelve weeks is comfortable. A rebrand or a channel switch in the middle of that history hurts more than a short one.
- Whether anything else is running. A second geo test at the same time, or a regional promotion, contaminates the read. One test at a time, nationally.
If the answer is no
There is no single volume that makes a geo test work, which is why the checker asks for two numbers rather than quoting you a threshold. A few hundred orders a week is around where results start to become useful, but whether yours do depends on the channel you want to test and how your sales sit across the country. Put your numbers in above and it will tell you.
If your total volume is fine but the channel is too small, test a bigger channel first, or group several into one test and measure them together. You lose the per-channel split, but you get an answer.
See the real numbers for your account
The checker works off bands. The designer works off your actual sales by region, and tells you which regions to hold out, how long to run, and what precision to expect. It is free and needs no account.