A two-box discount goes to every lapsed subscriber, and the quickest to take it are the quickest to leave again. Scored on a year it never saw, the targeted offer nets 24% more and gives away 64% less discount.
The retention lead decides which lapsed subscribers get the win-back offer of two boxes at half price. The rule is to send it to everyone who lapses. It feels fine because the offer only costs money when someone takes it, and the win-back report counts every return as a win. The report doesn't show who takes the two cheap boxes and leaves again.
We name the baseline. Offering everyone stays in every comparison, and the team's best simple cut, tenure before lapsing, is registered beside it.
We write down the expectation before the data is opened. The likeliest takers should keep the fewest boxes, and the score should beat the tenure cut by a stated margin.
We build on the first year of win-back sends. Each subscriber gets two estimates multiplied together: the chance they take the offer, and the full-price boxes they keep in the six months after it.
We freeze the score and write down a forecast of $58,900. Then we run it once on the second year's 18,400 lapsed subscribers. The result is $56,856, 3.5% under the forecast.
We try the simpler ways. Tenure, recency of lapse, and voluntary cancels only are each scored on the same unseen year. The score is 14% to 24% ahead of all of them and of offering everyone.
We hand back the score in the subscription platform's export, the line that turns it into a send list, the memo, and the holdout cell.
Subscribers below the line take the offer at 15.0% and keep 0.3 full-price boxes. Those above it take it at 12.5% and keep 4.1.
So what Rank on what a return is worth, not on the chance of a return.
The 11,040 subscribers below the line lose $10,930 between them on the judged year.
So what Stopping their offer is a gain as long as the discount never brings back someone who would later pay full price.
Offering only to subscribers with six months or more nets $49,870. The score is 14% ahead.
So what A simple cut recovers part of the gain, and the team can see exactly how much.
The team pulls the win-back list from the score each month and sends the discount only to subscribers whose expected return covers it. Everyone else gets a full-price note. A random 10% of the offer list, about 736 subscribers a year, gets the full-price note instead of the discount. Comparing that cell with the rest of the list measures how many returns the discount actually causes.
Some of the 920 who take the offer would have come back at full price. Only the holdout cell can say how many, and the $56,856 shrinks by that share.
Boxes kept past six months sit outside the window. A subscriber who stays two years is valued like one who stays six months.
The method transfers. The 40% line does not.
We'll tell you what your own history can say about it, and which offering fits. If none does, we'll say that too.
Start a conversation