What is a conversion lift study?

Google's help page describes Conversion Lift as a tool that measures the conversions that happen because people saw your ads, with test groups built either from user data or from geography¹. It also says the tool is not available for all accounts and points you to a Google representative¹. Meta's developer documentation says it randomly assigns people to a test or a control group, and that access is limited and goes through a Meta representative².

This makes it a close relative of a holdout test. The idea is the same. Some people do not see the ads, and the gap shows what the ads did.

The platform sells the ads. It also designs the test, runs it and counts the result. That does not make the result wrong. It does mean you should look at the set-up, not just the headline.

One AdExchanger article makes a related point: do not let the platform decide whether a sale was incremental, or whether a customer was new³. A customer who is new to the platform may already be in your own CRM.

Look for these before you act on a result:

• The size of the control group.
• The uncertainty around the lift, not only the lift itself.
• Which conversions were counted, and whether they match your own orders.

To turn lift into money, divide the extra revenue by spend. That gives iROAS, or incremental ROAS.