Marketing & Advertising
Loyalty Campaign Incrementality Calculator
Compare treatment and holdout purchase behavior, normalize the holdout to the treated population, estimate incremental revenue and contribution, and calculate cost per incremental purchase.
Treated members-
Holdout members-
Observed treatment purchases-
Observed holdout purchases-
Expected treatment purchases without campaign-
Incremental purchases-
Observed treatment revenue-
Counterfactual treatment revenue-
Incremental revenue-
Incentive, communication, and fixed cost-
Net incremental contribution-
Relative purchase-rate lift-
Cost per incremental purchase-
Decision view
Treatment, holdout, and counterfactual lift laboratory
Treatment, holdout, and counterfactual lift laboratoryPopulation-normalized purchase bars and a revenue bridge separate baseline demand, campaign lift, and the full cost of creating that lift.
Exact scenario comparisonTreatment purchase rate (%) changes while all other entered assumptions remain constant.
| Treatment purchase rate (%) | Treated members | Holdout members | Observed treatment purchases | Observed holdout purchases | Expected treatment purchases without campaign | Incremental purchases | Observed treatment revenue | Counterfactual treatment revenue | Incremental revenue | Incentive, communication, and fixed cost | Net incremental contribution | Relative purchase-rate lift | Cost per incremental purchase |
|---|
How to use Loyalty Campaign Incrementality Calculator
- Enter eligible members and the actual randomized treatment allocation.
- Use mature purchase rates and average purchaser spend from treatment and holdout.
- Include rewards paid on all treatment purchases, not only purchases labeled incremental.
Calculator guide
Understanding Loyalty Campaign Incrementality Calculator
Loyalty campaign revenue includes purchases that would have happened anyway. This calculator uses a randomized or defensible holdout to estimate the counterfactual behavior of treated members and isolates incremental purchases, incremental revenue, and net incremental contribution.
Detailed calculation process
Detailed loyalty treatment, counterfactual, and incrementality calculation
The default campaign assigns 80% of 50,000 eligible members to treatment and retains 20% as a holdout.
General formula: T=N*sH=N-TP_T=T*r_TP_H=H*r_HP_0=T*r_HDeltaP=P_T-P_0R_T=P_T*a_TR_0=P_0*a_HDeltaR=R_T-R_0C=P_T*i+T*m+FK=DeltaR*g-CLift=(r_T-r_H)/r_H The holdout rate and spend are projected onto the treatment population to estimate baseline demand. Only the difference from observed treatment revenue is considered incremental.
What each symbol means
Worked substitution with the default inputs
The default campaign creates an estimated 1,400 incremental purchases and $13,180 of net incremental contribution, subject to the validity of the holdout.
Worked situations
Practical examples
- The default treatment contains 40,000 members and 7,200 observed purchases; the holdout implies 5,800 purchases without treatment.
- Observed treatment revenue exceeds its scaled counterfactual by $161,000, creating $13,180 of net incremental contribution after all campaign costs.
Better inputs
Useful tips
- Assign the holdout before message exposure and preserve it through the purchase window.
- Use net merchandise revenue consistently in both groups.
- Check statistical uncertainty separately before acting on a small modeled lift.
Before relying on the result
Limitations and common mistakes
- This deterministic calculator does not calculate confidence intervals or correct non-random assignment.
- Spillover between treatment and holdout can bias the counterfactual.
- Long-term habit formation, pull-forward, and customer-level margin differences are not modeled.
Reference
Key terms
- Counterfactual
- The estimated treatment-group outcome had the campaign not occurred.
- Incremental purchase
- Observed treatment purchases above the population-normalized holdout expectation.
- Relative lift
- Treatment purchase-rate improvement divided by the holdout rate.
Important note
Do not call the output causal unless treatment assignment and measurement support that claim. Review sample balance and confidence intervals outside this deterministic estimate.
Frequently asked questions
Can I use last year's buyers as the control?
Only with strong caution; seasonality and selection make a contemporaneous randomized holdout preferable.
Why is incremental revenue not simply incremental purchases times treatment spend?
The model also recognizes the difference between treatment and holdout purchaser spend.
Does positive contribution prove significance?
No. Economic value and statistical confidence are separate tests.