Learn · Measurement
Understanding incrementality
Incrementality measures the extra revenue or behaviour your loyalty programme creates beyond what would have happened anyway. It compares programme members against a holdout group that receives no rewards. For a first programme, this isolates true impact from existing loyal customers, guides reward investment, and prevents paying for sales you would have earned for free.
How it works
Incrementality compares outcomes between two groups of customers who are alike in every way except exposure to the loyalty programme. You randomly assign a small holdout group before launch. This group never sees rewards, points, tier messaging, or programme offers. The main group receives the full programme. After a fixed period, you measure the difference in spend, purchase frequency, or retention between the groups. That difference is the incremental lift. Because both groups would otherwise behave similarly, any gap is attributable to the programme rather than to pre-existing loyalty.
A holdout group does not need to be large. For a first programme, five to ten percent of eligible members is enough to detect a material difference if the programme works. Keep the holdout assignment clean. Do not send programme emails, award points, or show member pricing to holdout customers. If a holdout customer buys through a partner that shares data with your loyalty register, treat that as baseline behaviour, not programme effect.
Worked example
This worked example is illustrative. Suppose you enrol 9,000 members in the programme and hold out 1,000. Over six months, programme members spend an average of 520 dollars each. Holdout members spend an average of 460 dollars each. Incremental spend per member is 60 dollars (520 minus 460), a lift of 13 percent (60 divided by 460, rounded to the nearest whole point). Total incremental revenue from the 9,000 programme members is 540,000 dollars (60 times 9,000). If programme operating cost is 180,000 dollars and rewards cost 2 percent of incremental revenue, reward cost is 10,800 dollars (540,000 times 0.02). Total cost is 190,800 dollars, so net incremental contribution is 349,200 dollars.
What to do
Before launch, define one primary metric such as incremental revenue per member or incremental purchase frequency. Select a random holdout group of five to ten percent of eligible customers using your loyalty register or CRM, and exclude that group from every programme touchpoint. Run the programme unchanged for at least one full purchase cycle, typically three to six months for most retail or service businesses. At the end, compare the holdout and programme groups on the primary metric. Calculate incremental lift and subtract programme costs including rewards, technology, and staff time. If the net result is positive, continue and test higher reward levels. If not, simplify the programme or adjust the reward threshold before scaling.
Common mistakes
- Measuring total member spend and calling it incremental, which ignores that many members would have bought at the same level without the programme.
- Allowing holdout customers to receive points through a partner or accidental email, which contaminates the control group and deflates your measured lift.
- Running the comparison for only a few weeks, so seasonal spikes or one-off promotions hide the true programme effect.
- Treating every new member's first purchase as fully incremental, when sign-up often coincides with an existing intent to buy.