What one point of churn is worth.

How this calculator works

Four numbers go in, and the model behind them is deliberately small — a small model you can check by hand beats a large one you have to trust.

Period churn is cancellations divided by the subscribers you started the period with. New subscribers added during the period are excluded from that denominator on purpose: someone who signed up on day 26 has had four days to cancel, not thirty, and folding them in makes churn look better than it is. The adds come back for the net-growth line, which is where they belong.

Monthly churn converts the period rate to a 30-day rate geometrically — one minus the survival rate raised to the power of thirty over the number of days in your period. Geometric rather than linear, because churn compounds against a base that is already shrinking.

Annualised churn applies the same compounding twelve times. This is the step that catches people out. Six and a half percent a month is not seventy-eight percent a year. It is closer to fifty-five, because each month’s churn only ever applies to the subscribers still there to leave.

The retention curve is that same expression read forward: survival rate raised to the number of months. Average lifetime is one divided by monthly churn, and lifetime value is revenue per subscriber times gross margin times lifetime. We use margin rather than top-line revenue because contribution is the only version of lifetime value you can actually spend.

Why one point is the number to argue about

Churn improvements are not linear in value, and that is the entire reason this calculator exists. Lifetime is the reciprocal of churn, so the same one-percentage-point move is worth more the lower your churn already is. Going from 8% to 7% adds about 1.8 months of lifetime. Going from 4% to 3% adds about 8.3 months. The work gets harder and the payoff gets larger at the same time.

One point off monthly churn, applied across a base you already own, is a capital-expenditure-sized number that arrives with no new acquisition spend attached to it.

Before you treat it as a retention problem

A single churn number hides the most useful split in the business. Somewhere between 20% and 40% of subscription churn is involuntary — a card that expired, a bank that declined a routine renewal, a billing profile that broke when someone moved house. Nobody in that group decided anything. They get counted as churn, they get treated as churn, and then a retention programme gets aimed at people whose only problem was a payment method.

Which half is bigger changes what you should build. If most of your churn is involuntary, the fix is retry logic, card-updater coverage and a dunning sequence that reaches people on more than one channel — see payment recovery. If most of it is voluntary, the fix is a cancel flow that captures a reason before it offers anything, pause and skip made visible rather than buried, and self-service quantity and cadence changes — see subscription retention. Building the second when you needed the first is the most expensive mistake in this category.

What good looks like

These are the published reference points we benchmark against. They are consumer-goods and DTC subscription figures rather than SaaS, which matters — SaaS churn benchmarks are far lower, and quoting them at a supplement or skincare brand is how founders end up believing their business is broken when it is average.

The retention curve is the part most worth putting in front of a board, because it turns a percentage into a calendar. At 6.5% monthly churn — the consumer goods benchmark — half a cohort is gone by roughly month ten. That is the window your onboarding, education and replenishment timing have to work inside, and it is a much shorter window than most brands plan for. If your post-purchase education sequence is scheduled to land at day 90, check the curve first and see how much of the cohort is still there to receive it.

Where this model stops being true

Three assumptions are doing real work here, and all three are wrong in known directions.

Constant hazard. The model assumes every subscriber is equally likely to leave in any month. Real subscription cohorts do not behave that way — churn is heaviest in the first two or three cycles and then flattens hard, because the people who were going to discover the product doesn’t suit them have already done it. So the calculator understates the lifetime of a subscriber who has survived six cycles, and overstates the lifetime of one you acquired last week.

One period is a small sample. If your window covers a promotion, a stockout, a price change or a delivery failure, the churn you have measured belongs to that event and not to your business. Run three or four consecutive periods before you believe a trend.

Blended cohorts. A single number averages your best acquisition channel with your worst, and your annual plans with your monthly ones. Annual billing roughly doubles twelve-month retention against monthly billing. If your billing mix is shifting, the blended rate will move without anything real having changed.

What to do with the result

1. Split the number. Pull cancellation reasons against dunning outcomes and get the real voluntary-versus-involuntary ratio instead of the benchmark band.
2. Re-run this by cohort — by acquisition channel, by first SKU, by billing interval. The blended rate tells you the size of the problem; the cohort rates tell you where it is.
3. Put the one-point figure next to the cost of the work. Cancel flows, retry logic and replenishment timing are fixed-cost builds against a return that scales with the base.
4. Give it a full purchase cycle before you read the result. For a monthly subscription that is 90 days minimum.