TLDR: Every churn tool tells you "it works." Almost none of them show you what actually worked, and for whom. Which save move, on which product, for which ICP, in which situation. That gap matters, because what rescues one customer does nothing for the next. So I built the answer into ChurnTools.
"It works" is not attribution
A green checkmark and a headline retention number feel reassuring and teach you nothing. If your churn dipped last month, which of the six things you changed caused it? For which segment? A tool that just says "it works" leaves you unable to double down, because you do not know what to double down on.
This is the same trap as reading a single churn rate. The aggregate hides the mechanism. And retention is all mechanism.
Why "what worked for you" is not "what works for them"
A discount saves a price-sensitive self-serve user and insults an enterprise buyer who was leaving over a missing integration. A pause option rescues a seasonal customer and does nothing for someone who never activated. The right save depends on the product, the ICP, and the specific reason that customer was walking. Attribution without those cuts is just a vanity metric wearing a lab coat.
What I built instead
The idea is simple: attribute the saves, not just count them. Tie each retained customer back to the intervention that actually moved them, and slice it by product, ICP, and churn reason, so you can see the pattern instead of guessing at it. That turns retention from "we tried some stuff and the number went down" into "this specific move works for this specific segment, do more of it."
You can see how it fits into a broader retention system across the experiments library and the tools directory. If you want a fast read on where your own biggest leak is before you start attributing saves, the Churn Health Check scores your setup in about 60 seconds, and ChurnTools is where the rest of it lives.