Discounts were recovering orders at a cost.
The beauty retailer’s abandoned-cart sequence escalated toward discounts and regularly claimed recovered revenue. Those customers already had high buying intent, so attributed sales did not show how many additional purchases the emails created or whether the incentive was necessary.
Count orders and the margin they leave
Over ten weeks, incremental recovered orders increased 18%, discount use fell 26%, and recovered-order margin after variable costs improved by five percentage points. The comparison supports the recovery design as a whole; it does not identify an independent causal effect for every message change.
Address the hesitation before reaching for a discount
Cart messages answered likely buying concerns through reassurance and product information. Incentives stayed selective, with their cost visible in the recovered order.
A randomly selected group receiving no campaign helped estimate additional purchases. The review combined those orders with discount use and margin after variable costs, so higher revenue could not conceal weaker margin.
Recover the purchase, not just the tracking credit.
The sequence addressed buying concerns before offering an incentive. A group receiving no recovery campaign helped estimate additional orders and the margin left after discounts.
- Natural recovery
- Crediting all purchases after an email as additional orders created by that email.
- Incentive cost
- Treating a discounted recovered order as economically equal to a full-price order.
- Relevant reassurance
- Escalating discounts before addressing a purchase concern that information could resolve.
- Contribution after recovery
- Calling revenue growth successful when the additional orders retain insufficient value.
An attributed order might have happened anyway. The decision followed added orders and retained margin.
Measure purchases added by the campaign
A control group carries an opportunity cost because some uncontacted customers may have responded to a message. That cost had to be proportionate to the value of resolving whether the sequence was adding profitable orders.
Discount reduction was also not an isolated target. Incentives could still be useful where justified, and the relevant outcome was the contribution left after recovery rather than the most impressive email-platform revenue total.
The recovery sequence had to earn additional contribution. That meant measuring orders beyond natural return and checking the margin left after incentives, rather than celebrating revenue attributed to the last email.