Subscription churn rate: how to calculate it and how to cut it

Monthly churn rate = subscribers lost during the month ÷ subscribers at the start of the month. Leave out people who signed up during the month, or a strong acquisition month will make churn look better than it was.

Monthly and annual churn don't convert by multiplying by 12, because each month's churn applies to a smaller base. Annual churn = 1 − (1 − monthly churn)¹². At 2% a month that's about 22% a year, not 24%. At 6% a month it's 52%.

The average subscriber lifetime is roughly 1 ÷ monthly churn. At 5% a month, the average subscriber stays about 20 months. Cutting churn to 4% stretches that to 25 months, a quarter more revenue from every subscriber you already paid to acquire.

Revenue churn uses the same formula with recurring revenue instead of subscriber counts. It's worth tracking alongside it if customers often downgrade, skip, or move to cheaper plans.

Published churn benchmarks are thin and inconsistent, so treat any single number with care. Recurly's network data from July 2026 puts ecommerce, meaning subscription boxes and direct-to-consumer brands, at 4.25% median churn: 2.87% voluntary and 1.38% involuntary. That's the second-highest of the industries it tracks, after education. Recurly labels those figures annual, though at that level they read more like monthly rates, so check which period you're comparing.

The split is the more useful part, and it holds whichever period it is: about 32% of ecommerce churn is involuntary. In that data, roughly one in three subscribers a DTC brand loses didn't decide to leave. Recurly's data also shows involuntary churn falling sharply as price rises, from 1.30% for subscriptions at $10 to $25 a month to 0.18% above $250.

The best comparison is your own churn by cohort: subscribers who joined in the same month, tracked month by month. If newer cohorts churn faster than older ones did at the same age, something changed, whether it's the offer, the channel, or the product.

Involuntary churn is the cheaper half to fix, because the customer still wants the product. The fixes run without anyone touching them:

• Card updaters and network tokens, so expired and reissued cards renew themselves.
• Retries timed by decline type, inside the card network limits.
• A no-login link to update the card, sent before the subscription lapses.

The full playbook is in recovering failed subscription payments, and the decline codes reference covers which failures are worth retrying.

The common reasons people cancel each have something you can offer at the moment they try:

• Too much product piling up. Offer a skip, a pause, or a longer gap between orders.
• Bored of the flavor, scent, or size. Offer a swap.
• Price. Offer a smaller or cheaper version before a discount.
• Moving, traveling, or a short-term budget squeeze. Offer a pause with a date it resumes.

A cancellation flow that asks why and responds to the answer tends to save more subscribers than a single discount shown to everyone. Customers who leave anyway are worth a win-back message a few weeks later, once the reason they left may no longer apply.

Failed renewals are one of four places a checkout loses money, alongside declined orders, chargebacks, and processing fees. The Payments Leak Calculator puts a yearly number on all four.

Find your payments leak