A per-customer next-best-offer, drawn from the offers you already run. Ten illustrative rows from the 50,000 scored — your full preview file contains every row.
| Customer | Signal read | Next best offer | Predicted uplift | Action |
|---|---|---|---|---|
| a91f…3c | Lapsing · price-sensitive | Restock bundle, no discount | +8.4% | Contact |
| 7be2…9d | Active · high frequency | Early access to new drop | +6.1% | Contact |
| c044…1a | Dormant · high past value | Win-back, free delivery | +11.7% | Contact |
| f3a8…b2 | Recent buyer · satisfied | — hold, likely to buy anyway | no contact needed | Suppress |
| 9d12…7e | Lapsing · category-loyal | Replenishment reminder | +9.9% | Contact |
| 2bb0…4f | Discount-trained | Loyalty points boost, no markdown | +5.3% | Contact |
| 8c71…0a | One-time buyer · low intent | — hold, low expected return | below threshold | Suppress |
| d4e9…11 | Active · gift-buyer | Cross-sell complementary line | +7.6% | Contact |
| 5af3…c8 | Lapsing · premium tier | VIP early access, no discount | +13.1% | Contact |
| b6d2…9f | Seasonal buyer | Pre-season teaser | +6.8% | Contact |
Extrapolated from this segment's scoring to your active base. A projection, not a measured result — the measured number comes from the full pilot's holdout test.
Estimated incremental revenue if the same approach is applied across your active base, versus your current business-as-usual send.
How we get there: +9.2% average predicted uplift per contacted customer, applied to your contactable base, net of the 31% we'd suppress to avoid wasted discounts.
Range reflects modelling confidence. Verified in the full pilot against a matched holdout.
This preview shows the recommendation quality. The paid pilot turns it into a number your finance team can sign off.
The full NBO file for your entire base — not just this 50k slice — with fallback offers and send timing per customer.
A statistically matched control group so the result is measured, not projected.
Actual incremental revenue per contact after the campaign, with confidence intervals and discount spend avoided.
Where you're currently discounting customers who'd have bought anyway — and what stopping is worth.
The full measured pilot is $3–6k by customer count — credited against year one and refunded if it doesn't beat your baseline. You only pay now that you've seen it work on your own customers.