Measurement
Holdout group
A group of people deliberately not shown the ads, used as the comparison in a test.
What it is
Because the group is chosen at random, its behaviour shows what would have happened without the ads.
Why it matters
For lead generation and e-commerceIt is the basis of every honest measure of what ads cause.
In practice
Hold out a random share of the audience or of regions, run the campaign as usual, and compare results.
A common mistake
Holding out a group that differs from the exposed one, such as a different region or season, which invalidates the comparison.
All terms