Incrementality tests
Test where the spend is big, you are least sure of it and a budget decision is pending. Judge every test on total sales or profit, not on what a platform or a code reports.
7 answers · Data as of Nov 22, 2026

The short version
Our judgment, drawn from the 7 answers below.
Test first the big or growing spend you are least sure of, and skip formal tests below about $40K a month per channel.
From Which channels to test first, Incrementality test cadence across channels
Set a geo holdout's pass bar at breakeven, and use 12 or more regions per group, at least 15% held out and 4 weeks or more.
Hold out a random share of customers from email, SMS or costly discounts, and judge revenue or profit per person, not redemptions.
From Testing whether email and SMS add sales, Testing discounts for incrementality
Use a platform's free lift study for questions inside that platform once you qualify, and an independent geo test for comparisons across channels and big budget moves.
The 7 answers
Grouped by the decision you are making. Jump to a group:
Pick what to test
How often to test, what to test first and which kind of test to use.
How often should a brand our size run incrementality tests?
Plan incrementality tests every quarter once a channel can be read. Test where a budget decision is pending or evidence is oldest, and retest after big changes. Below about $40K a month per channel, skip formal tests.
Medium confidence · 15 sources
Which channels should we test first for incrementality?
Test first the big or growing spend you are least sure of, with a budget decision pending. That is usually your largest channel, or branded search and retargeting when they carry real spend.
Medium confidence · 17 sources
Should we use platform lift studies (Meta, Google) or independent geo tests?
Use a platform's free lift study for questions inside that platform once you qualify. Use an independent geo test for sales it can't see, comparisons across channels and big budget moves. Check one against the other now and then.
Medium confidence · 16 sources
Design the test
How to set up geo, email, discount and national tests.
How should we design a geo holdout test for our order volume?
Set the pass bar at breakeven, then size the test on your daily sales by region. Use 12 or more regions per group, hold out at least 15%, run four weeks or more and measure total sales.
Medium confidence · 17 sources
How should we test whether email and SMS campaigns add sales?
Hold out a random 5 to 20% of the eligible audience from one email or SMS program for at least 2 to 4 weeks. Then compare store revenue per person between the groups, net of discounts.
Medium confidence · 10 sources
Should we test discount codes and promotions for incrementality?
Yes. Test every discount you run often or that costs real margin, starting with the costliest. Withhold it from a random 10 to 20% of eligible customers and judge profit per customer, not redemptions.
Medium confidence · 13 sources
How do we measure the added sales from a channel or offer we can't hold out by region or customer, such as national TV or a sitewide sale?
Run the buy in a fixed set of markets first and read total sales against a synthetic control. When it reaches every market at once, forecast sales from your own history and treat the gap as a rough read.
Medium confidence · 9 sources
More in Measurement
69 answers in 10 subjects
Guides that use these answers: How to tell if your ad spend is working · How to add a second ad channel