You can pour new customers into the top of a subscription business all day. If they drain out of the bottom at the same rate, you run very fast and stay exactly where you are. That drain is churn, and the reason it kills quietly is that the percentages sound harmless. Five percent a month is a rounding error in conversation and a catastrophe in a spreadsheet.
Two churn rates, and why you need both
Customer churn = customers lost during the period ÷ customers at the start of the period × 100. Start January with 200 customers, lose 10, and you have 5% monthly customer churn.
Revenue churn = MRR lost during the period ÷ MRR at the start × 100. This is the more important one, because customers are not interchangeable. Losing ten customers on your $19.99 plan and losing three on your $69.99 plan are similar customer churn events and very different revenue events.
The gap between the two numbers is diagnostic. If revenue churn is much higher than customer churn, you are losing your best customers, which is the more dangerous situation. If revenue churn is much lower, you are losing small accounts, which usually means a fit problem at the entry tier rather than a product problem.
One refinement worth knowing: net revenue retention includes expansion. If existing customers upgrading adds more MRR than cancellations remove, net revenue retention is above 100%, and your revenue grows even with zero new customers. That is the strongest position a subscription business can be in, and it is entirely possible for a business with visible churn to be in it.
The compounding, in real numbers
Churn is monthly and multiplicative, which is why intuition fails. At 5% monthly churn, retention after a year is 0.95 to the power of 12, which is about 54%. You lose nearly half your customer base every year just standing still, and all your acquisition effort in the first six months goes to replacing people rather than growing.
At 3% monthly, annual retention is about 69%. At 2%, about 78%. Those look like small differences on a monthly line and they are enormous over two years. The rough average customer lifespan is 1 divided by the monthly churn rate: 20 months at 5%, 33 months at 3%, 50 months at 2%. Since customer lifetime value scales directly with lifespan, cutting churn from 5% to 3% raises the value of every customer you will ever acquire by roughly two thirds, without changing your price or your acquisition at all.
That is the argument for treating churn work as growth work. It is usually cheaper to keep a customer than to find a new one, and the effect applies retroactively to your entire base.
Involuntary churn is the cheapest thing to fix
A meaningful share of cancellations are not decisions. They are expired cards, insufficient funds, bank fraud blocks, and 3D Secure prompts nobody responded to. The customer still wants your product and does not know they have been cut off. This is involuntary churn, and it is the highest-return churn work available because there is no persuasion involved.
- Turn on smart retries in your payment processor so failed charges are attempted again on a sensible schedule rather than immediately.
- Enable the card account updater feature, which refreshes expired card details automatically for many card networks.
- Send a dunning email sequence that actually explains the problem and links straight to a card update page.
- Notify customers before an annual renewal charges, so a surprise does not turn into a chargeback.
- Watch for cards expiring in the next 30 days and prompt those customers ahead of the failure.
Most payment platforms, Stripe included, provide these as configuration rather than engineering. Setting them up is an afternoon and it recovers revenue every month afterwards.
Voluntary churn is decided in the first week
For customers who do choose to leave, the decision is usually made far earlier than the cancellation. If someone signed up, did not reach the moment where the product became useful, and drifted away, the cancellation two months later is a formality. Fixing that means fixing onboarding, not writing a better retention email.
Identify the specific action that correlates with people staying. For a scheduler that might be connecting a second account and scheduling a first post. For an analytics product it might be connecting a data source. Whatever it is, that action is your activation event, and the job of onboarding is to get people to it in the first session. Everything else in the onboarding flow can be cut.
Then look at where people stall. Session recordings and a simple funnel from signup to activation will usually show one step where half the users disappear. That step is worth more attention than any feature on your roadmap.
Ask people why, and actually read the answers
Put a short, optional question in the cancellation flow. Not a survey, one question with four or five options and a free-text box. The answers cluster fast, and the clusters tell you what kind of problem you have.
“Too expensive” usually means the value was not obvious, not that the price is wrong. “I did not use it” is an onboarding failure. “Missing a feature” is real product feedback, and worth acting on only if it repeats. “Project ended” is not churn you can fix and should be excluded mentally from your improvement efforts.
Tag each cancellation with the channel that acquired the customer while you are at it. If “I did not use it” clusters heavily among people who arrived from one particular platform, you have found a marketing problem wearing a product costume. Tools that keep acquisition source attached to the Stripe customer record, which is what seenpaid does read-only, make that cross-reference a filter rather than a research project.
Read the free text yourself, in full, for the first few hundred cancellations. There is no dashboard that substitutes for this, and the phrasing people use is often the phrasing that belongs on your landing page.
The churn that starts with your marketing
The cause of churn that gets the least attention is acquisition. Customers who were never a fit churn hardest and fastest, and they arrive because something about your marketing promised them a different product than the one you built.
This happens most often with content that performs well for the wrong reasons. A post that goes wide attracts a broad audience, some of whom sign up because it was entertaining rather than because they have the problem you solve. They convert, they churn in six weeks, and your acquisition metrics look great while your retention quietly degrades. Meanwhile the narrow, technical post that brought in eleven signups produced customers who are still paying two years later.
You cannot see that pattern from a channel report that stops at signup. You need to follow specific cohorts from the post that acquired them through to whether they stayed. When you do, the recommendation is usually to write more of the content that performs worse on every social metric, which is an uncomfortable but profitable conclusion. seenpaid connects read-only to Stripe and attributes revenue to individual posts over time, which lets you separate the channels that bring keepers from the ones that bring quitters before churn becomes the number defining your business.