Aashna Jain

What is involuntary churn, and why do subscriptions die at the payment step?

3 min readFor buildersBuilding on it21 of 21

A large share of the customers you think you lost never decided to leave. Their card expired.

Two ways to lose a customer
subscribercancelledthey decidedfailednobody decidedanythingthis is the recoverable one

Only one of these is feedback about the product.

Where the failures come from

Four causes account for most of it.

Expired cards. Cards expire on a schedule and customers do not update them until something stops working. Network account updater services fix a meaningful share of this automatically, and many teams have simply never enabled them.

Insufficient funds. Timing rather than intent. The same card charged four days later frequently succeeds, which is why retry scheduling matters more than retry count.

Issuer risk rules. A recurring charge that looks unusual, arrives from a new descriptor, or trips a velocity rule gets declined even though the money and the card are both fine.

Authentication. In markets requiring strong customer authentication, some recurring charges need the customer present, and a renewal at three in the morning does not have one. Correctly flagged merchant-initiated transactions avoid this. Incorrectly flagged ones fail silently and repeatedly.

Why it is worse than it looks

Because it is invisible in the metric that gets reported.

Total churn blends two completely different things: people who evaluated your product and left, and people who never had the conversation. Averaged together, they produce a number that suggests a product problem when part of it is a billing problem, and teams respond by changing the product.

It also compounds. A customer who is cut off, notices a week later, and has to re-enter payment details has been given a natural moment to reconsider the subscription. The failed payment did not just cost you a month. It handed the customer a decision point they were not going to reach on their own.

A gym where the door code silently stops working on some members' cards. They turn up, cannot get in, go home, and eventually stop turning up. The gym reviews its class timetable, because the members did not cancel and nobody wrote down why they stopped coming.

What actually fixes it

Separate the metric first. Report voluntary and involuntary churn as two numbers. Nothing else on this list gets funded until somebody sees the split.

Retry on a schedule shaped by the decline reason, not a fixed interval for everything. Insufficient funds wants a few days and a date near payday. A do-not-honour code wants a different treatment entirely, and a hard decline wants none.

Turn on account updater with both major schemes.

Dun like a human. Tell the customer before access ends, not after, and make updating the card a two-tap job rather than a login-and-navigate job.

Flag transaction types correctly. Merchant-initiated versus customer-initiated is not paperwork. It decides whether a renewal needs a customer who is not there.

Where you meet it

Every subscription that quietly stopped working. Every "update your payment details" email that arrived after access had already gone. Every product team convinced it has a retention problem.

Building this? A second pair of eyes on the architecture is what the advisory is for. →

Get the next one in your inbox.

New fundamentals and one memo every two weeks.

One memo every two weeks. Unsubscribe in one click. Delivered through Substack, so their terms and privacy policy apply.