Control group in marketing: what it is and how to use it to measure campaigns
FLYDE Glossary · November 9, 2026 · 4 min read
A control group is a randomly chosen part of a campaign's audience that does not receive the campaign. Comparing how much it buys with the group that does tells you how many sales the campaign really caused.
Why you need a control group
Many customers who buy after receiving a campaign would have bought anyway. Without a control group, all those purchases are credited to the campaign and its result is inflated. With one, the difference between the two groups is the incremental sales: the sales that would not have existed without the campaign.
How the result (lift) is calculated
An example. You send a reactivation campaign to 20,000 customers and hold out 2,000 chosen at random. 6% of those who received it buy, against 4% of the control group. Lift is (6 − 4) ÷ 4 = +50%, and the sales the campaign can claim are those in the 2-point difference, not the full 6%.
Calculate your campaign's lift
How to build a good control group
- Random and before launch. If you choose the group afterwards or by criteria, the two groups are no longer comparable.
- The same relative size in every segment, so results can be compared by segment.
- Large enough. 5% to 20% of the audience is typical; the smaller the difference you want to detect, the bigger the group must be.
- The same across all channels. If the control skips the email but sees the ad, it is no longer a control.
- Measured on real sales in store and online, not on the clicks or conversions each platform reports.
Per-campaign or permanent control group
A per-campaign control group measures one specific action. A permanent control group (or holdout) keeps part of your customers out of all communication for months and measures the cumulative effect of the marketing programme as a whole. They complement each other.
Control groups in FLYDE
In FLYDE the control group is set when you build the audience and is shared across every channel where you activate it. Results are matched against real ecommerce, store and CRM sales, and the incrementality module gives you lift, incremental revenue, incremental ROAS, margin and confidence interval.
Frequently asked questions
What is a control group in marketing? A randomly chosen part of a campaign's audience that does not receive it. It is the baseline for measuring how many sales the campaign really generated.
How big should a control group be? Typically 5% to 20% of the audience. It must be large enough for the difference between groups to be statistically reliable; with small audiences, use a higher percentage.
What is lift? The relative increase in conversion of the group that receives the campaign compared with the control group. A 50% lift means the exposed group converts 50% more than the control.
Don't I lose sales by not reaching the control group? You lose only the incremental sales of that small group, and in return you know which campaigns work. Without that measurement it is easy to keep investing in actions that sell nothing that would not have sold anyway.