An experimental design used to measure affiliate channel incrementality by comparing purchase behavior in groups exposed vs. not exposed to affiliate promotion. Methodology: select a control group (customers or geographic markets not exposed to affiliate links) and a test group (customers or markets with normal affiliate program exposure); run the test for 4-8 weeks to accumulate statistically significant data; compare the purchase rate (or conversion rate) of the holdout group against the test group; the difference represents the incremental lift attributable to the affiliate channel. Implementation approaches: geographic holdout: pause the affiliate program in selected geographic markets (e.g., 20-30% of markets) while maintaining it in others; compare purchase volume in paused vs. active markets controlling for baseline differences; customer-level holdout: using a cookie or device-based method to suppress affiliate link delivery to a random sample of site visitors; compare conversion rates between the suppressed group and normal visitors. Limitations: geographic markets have inherent differences that can confound results; customer-level holdout requires infrastructure that most brands don't have in standard affiliate network setups; holdout tests typically measure channel-level incrementality (does the affiliate channel as a whole add value?) rather than publisher-level incrementality (which specific publishers are incremental?). Use case: most valuable for brands questioning whether their coupon/cashback-heavy affiliate program is driving genuine new revenue or primarily capturing commission on sales that would have occurred through direct and organic channels.
Related Resources