ShopNow is deciding between two customer targeting strategies for a retention campaign: Strategy A, a propensity model that ranks users by predicted purchase likelihood, and Strategy B, a rule-based segment built from recent browsing and cart activity. The marketing team can contact only the top 20% of 1,000,000 eligible users due to budget limits.
Overall historical conversion rate in the evaluation set is 5.0%. Both strategies were evaluated on the same 1,000,000-user holdout sample.
| Segment | Users Targeted | Strategy A Conversion Rate | Strategy A Lift | Strategy B Conversion Rate | Strategy B Lift |
|---|---|---|---|---|---|
| Top 10% | 100,000 | 15.0% | 3.0x | 12.0% | 2.4x |
| Top 20% | 200,000 | 12.0% | 2.4x | 10.0% | 2.0x |
| Top 30% | 300,000 | 9.5% | 1.9x | 8.8% | 1.76x |
| Top 50% | 500,000 | 7.0% | 1.4x | 7.2% | 1.44x |
| Full Population | 1,000,000 | 5.0% | 1.0x | 5.0% | 1.0x |
Additional campaign economics:
| Metric | Value |
|---|---|
| Cost per contact | $0.40 |
| Gross profit per conversion | $18 |
| Max contact volume | 200,000 users |
The VP of Marketing wants to know which strategy should be used for the next campaign and how a lift chart should guide that decision, rather than relying only on overall model quality.