Subscription churn: how to measure it and how to reduce it

A product subscription promises predictable revenue. Churn is what makes it unpredictable. A store with a hundred subscribers and 10% monthly churn loses half of them in six months and has to keep pouring in new ones just to stand still. At 5% churn the same store is twice as big a year later — on the same ad budget. The gap between those two numbers decides whether the subscription makes sense at all.

The formula

Churn rate for a period is the share of subscribers who cancelled, out of those active at the start of the period:

Churn = Cancelled during the month ÷ Active at the start of the month

200 active subscriptions at the start of September, 14 cancellations during September — churn is 7%. New subscribers who joined and cancelled in the same month are not in the denominator; they are counted separately, more on that below.

The inverse is retention, 93% in the example. And the average subscription lifetime is roughly 1 ÷ churn: at 7% a month a customer stays about 14 months, at 10% — 10, at 15% — under 7. That lifetime is what multiplies LTV, and it is what we plugged into the LTV calculation: the difference between 7% and 15% is the difference between 14 and 7 orders per customer.

How product churn differs from SaaS

The numbers written about churn are almost always SaaS numbers: “a healthy level is 3–5% a month”. For a physical product subscription they do not transfer without adjustments.

A pause is not a cancellation. A coffee subscriber went on holiday and paused for a month. SaaS has no such state. Counting a pause as churn inflates the figure; ignoring it understates it, because some pauses never resume. The working rule: a pause longer than two consecutive delivery cycles counts as churn.

Product accumulates. A SaaS customer cannot “stock up” on the service. A product customer can: if the delivery frequency exceeds real consumption, coffee piles up on the shelf and the customer cancels not because they are unhappy but because they have nowhere to put it. This is the most common cause of churn in the first two months, and it has nothing to do with product quality.

A failed charge is not the customer’s decision. An expired card or no money on the day of the charge produces “involuntary” churn: the customer was not planning to leave, but the subscription stopped. In product subscriptions this is 20–40% of all losses, and it is the cheapest part of churn to fix.

So one churn figure says almost nothing. You need three:

What counts as normal

For subscriptions to products with a short consumption cycle (coffee, pet food, water):

• Voluntary churn of 3–6% a month is good. Up to 10% is tolerable at launch. Consistently above 10% is a frequency or product problem.
• Involuntary churn under 2% — given retries and card-update reminders. Without them it easily reaches 5–8%.
• Early churn up to 15% is normal. Some people try a subscription as a way to get a discount on one order; you will not eliminate them entirely.

More important than the absolute figure is the trend by cohort: do September’s subscribers cancel less than July’s? If so, your changes are working.

Reasons for cancelling and what to do about each

The most useful thing you can do about churn is one question on the cancellation screen: “Why?” with four or five options. Not to talk them out of it, but to know. Here is what people usually answer and what to do with it.

“The product is piling up”

Frequency exceeds consumption. This is not a reason for a discount — it is a reason to offer changing the frequency or skipping the next delivery right there on the cancellation screen. Stores that add a “skip” button next to “cancel” recover 30–50% of cancellations for this reason.

Prevention: the default frequency must match the real cycle, not the desired sales volume. A 250 g bag of coffee for one person is three weeks, not two.

“I want to try something else”

A subscription to the same coffee or the same pet food flavour gets boring. The fix is swapping the product inside the subscription without cancelling: the customer changes the origin before the next delivery, the subscription lives on. If the account does not allow it, the customer cancels and maybe starts a new one — or maybe not.

For coffee and cosmetics, rotation works on its own: “Ethiopia this month, Colombia next”. A subscription with built-in novelty lives longer than a subscription to the same thing.

“Too expensive”

The least informative answer — it comes from people for whom it really is expensive and from people who simply found it cheaper elsewhere. A discount at cancellation is a bad habit: customers quickly learn that clicking “cancel” gets them 15% off.

Instead, check two things. Whether the subscription’s benefit versus a one-off purchase is visible on every delivery (in the email: “you saved 8 versus the one-off price”). And whether your product has become more expensive than competitors’ for a one-off purchase — a subscription will not hold a customer at a price above the market.

“I can’t keep up” / “Circumstances changed”

A move, a holiday, a temporary change of habits. The only thing that works here is a pause with a resume date and a reminder a week before it. Half of those who would have cancelled choose the pause, and half of those come back.

No answer — the charge failed

Involuntary churn. This needs no psychology, it needs a process:

1. A retry after 1–2 days, then another after 3–5 days. A large share of declines are temporary.
2. An email after the first failure: “we couldn’t charge your card, please update it” with a link to the account.
3. An email 7 days before the card expires — before the charge fails.
4. The subscription is not cancelled automatically after the first failure. Only after 3–4 attempts and two emails.

How retries and notifications work with Ukrainian payment providers is covered in the article on recurring payments. If you do not have this in place, it is the first thing to do — cheaper than any work on voluntary churn.

Early churn: the first 30 days

Early churn is a separate story, because its cause is almost always the same: the first delivery did not match expectations. Too big, too early, the wrong grind, the wrong size.

What works:

• The first delivery is the same as the following ones. Not a month-long “starter kit” after which the second bag arrives before the first is opened.
• An email after the first delivery: “everything right? change the frequency / grind / quantity here”. One click and the subscription is adjusted to reality instead of being cancelled.
• The subscription discount should not exceed the discount a customer can get any other way. Otherwise the subscription is “a way to get 20% off one order”.

What to measure monthly

The minimum set, which fits in one table:

All six are ordinary reports from a merchant dashboard, if it provides them. Neocarts has them in analytics from day one; what it looks like for the customer and the merchant is on the recurring payments page, and the price is 5% on subscription transactions with no monthly fee.

The point

Churn is not the enemy of subscriptions; it is their main metric. A store that knows its churn by three types and sees it by cohort can manage it: pause instead of cancel, swap instead of cancel, retries instead of a quiet death. A store that knows one figure, “10% a month”, can only complain about it.

If you are still planning the launch, the order of decisions that reduces churn before the first delivery is in the step-by-step guide.