What is a geo experiment?
A geo experiment is a test that changes advertising in some geographic regions while keeping it the same in others, then compares results between the two sets of regions to measure the ads' incremental effect.
Last reviewed .
How a geo experiment works
You split markets, such as states or metro areas, into test and control groups. In the test regions you run, raise, or pause ads, and in the control regions you leave things as they were. After the test period, you compare outcomes like sales or sign-ups.
Matching regions on past performance helps the groups be comparable.
When to use one
Geo experiments suit channels that can't randomize individual people, such as TV, radio, out-of-home, and some digital campaigns. They also work without user-level tracking, which makes them useful in a privacy-limited setting.
They need enough regions, enough time, and results that can be measured by region. Spillover between regions, such as people traveling, can blur the result.
Common questions
How long should a geo experiment run?
Long enough to capture a full buying cycle and enough data to see the effect. Many run for several weeks, but the right length depends on your business.
How many regions do you need?
More regions give more reliable results, because they reduce the chance that one unusual market skews the comparison. There's no universal number, and a power analysis can estimate what you need.
Is a geo experiment the same as a holdout test?
A geo experiment is one kind of holdout test, where the held-out group is a set of regions. Holdouts can also be made up of individual users.
What can go wrong with a geo experiment?
Regions that aren't truly comparable, outside events in only some regions, and ad spillover across borders can all distort results.
Want help putting this to work?
Tell us where your marketing stands today, and we'll show you what to do first.