Subscriptions
Churn rate
Churn rate is the share of subscribers or subscription revenue lost over a period, most commonly reported monthly for DTC subscription businesses.
Blended DTC subscription churn ran 6.5–8.5% monthly in 2026, with 5–7% considered good and below 5% top-quartile. Category variance is wide: supplements and consumables 5–8%, meal kits 8–15%, general subscription boxes 10–15%.
A single blended number is a poor management metric because it averages cohorts that behave nothing alike — a 12–30% first-month spike and a much lower steady-state tail combine into a figure describing neither. Track cohort retention curves and the voluntary/involuntary split alongside it.
The denominator is the whole argument
Churn rate has an agreed numerator and no agreed denominator. Everyone counts the subscribers lost during the period. The disagreement is about what to divide them by, and the available choices produce materially different numbers from identical facts.
Take a month that opens with 10,000 active subscribers, adds 1,000 and loses 600. Four conventions are all in common use, and the results below span 5.71% to 8.00% on that one set of facts.
The last row is the one that matters most in DTC and is used least. With mixed cadences — monthly, every 60 days, prepaid six-month terms — a large share of the base had no renewal in the period and could not have churned at all. Dividing by everyone dilutes the rate by the share of the book that was never asked.
| Convention | Denominator | Churn rate | The argument for it |
|---|---|---|---|
| Opening base | 10,000 | 6.00% | Simple, stable, and every one of them was exposed to the risk |
| Average of opening and closing | 10,200 | 5.88% | New subscribers joined mid-month and were only partly exposed |
| Opening plus half of new | 10,500 | 5.71% | Approximates mid-month arrival without a daily calculation |
| Renewal-exposed base | 7,500 | 8.00% | Only subscribers with a renewal due could have churned |
Monthly and annual churn do not convert by multiplying
Churn compounds against a shrinking base, so annualising a monthly rate means retaining repeatedly rather than losing repeatedly. Annual retention is one minus the monthly churn rate, raised to the twelfth power; annual churn is one minus that.
The naive conversion — monthly rate times twelve — overstates at every level and becomes nonsense at the top of the range, where it returns figures above 100%. A business cannot lose 120% of its subscribers.
The conversion runs the other way too, and is needed more often than people expect, because subscription benchmarks are quoted monthly while board reporting is annual. A 50% annual churn rate is a monthly rate of one minus 0.5 to the power of one twelfth, or about 5.6% — not the 4.2% that dividing by twelve produces.
| Monthly churn | Annual retention | Annual churn | Naive 12x figure |
|---|---|---|---|
| 3% | 69.4% | 30.6% | 36% |
| 5% | 54.0% | 46.0% | 60% |
| 6% | 47.6% | 52.4% | 72% |
| 8% | 36.8% | 63.2% | 96% |
| 10% | 28.2% | 71.8% | 120% |
Customer churn and revenue churn are different rates
Losing 6% of subscribers is not losing 6% of subscription revenue, because the subscribers who leave are rarely average. In DTC they skew towards the cheapest plan, the longest cadence, a single product line and the deepest introductory discount.
Continuing the example: 600 subscribers lost from a base of 10,000, where the base bills $45 a month on average and the churners were billing $32. Customer churn is 600 divided by 10,000, or 6.00%. Revenue churn is 600 times $32 divided by 10,000 times $45 — $19,200 of $450,000, or 4.27%.
Both are true and they answer different questions. Customer churn describes the product experience; revenue churn describes what happened to the book. If revenue churn is the higher of the two, the customers leaving are the valuable ones, and that is by far the more urgent of the two findings.
Why one business reports three different churn numbers
The three numbers are almost never an error. They are three defensible definitions produced by three teams answering three different questions, and the confusion comes from publishing them without labels.
The fix is procedural rather than analytical. Name one definition as the reported churn rate, write it down where the number is displayed, and label every other version with the question it answers. A business arguing about which churn number is correct is usually arguing about which question it is asking.
These are the divergences worth labelling, any one of which explains a disagreement on its own.
- Denominator: opening base, average base, opening plus half of new, or renewal-exposed base. Worth more than two percentage points on the example above.
- Unit: customers or revenue, and if revenue, gross or net of expansion.
- Period basis: monthly, annualised by compounding, annualised by multiplication, or per renewal cycle — the only frame that means anything on a 60-day or prepaid cadence.
- Booking date: a cancellation requested on 3 March and effective at term end on 30 April is churn in March for the retention team and April for finance.
- Population: whether paused subscribers, subscriptions in failed-payment recovery and trials sit in the base, in the numerator, or outside both.
Frequently asked questions
- How do you calculate churn rate?
- Divide the subscribers lost in a period by a base, and state which base. Opening base is the most common: 600 lost from 10,000 opening subscribers is 6.00%. The average base, opening-plus-half-of-new, and the base actually exposed to a renewal give 5.88%, 5.71% and — if only 7,500 had a renewal due — 8.00% on the same facts.
- What is the difference between churn rate and retention rate?
- Within one period they are complements: retention rate is one minus churn rate for the same base and window. Across periods they stop being interchangeable, because churn compounds. A 6% monthly churn rate is not 72% annual churn; twelve months of 94% retention compounds to about 47.6% retained, so 52.4% annual churn.
- Is churn rate monthly or annual?
- Both are used, and they must not be converted by multiplying. Annual retention is monthly retention compounded twelve times: 6% monthly churn leaves about 47.6% retained after a year, so 52.4% annual churn rather than 72%. Going the other way, 50% annual churn is about 5.6% monthly, not 4.2%.
- What is a good churn rate for a subscription business?
- Category benchmarks exist, but they only compare when the definitions match — the same business can report 5.71% or 8.00% on identical facts depending on the denominator. Compare against your own trend and against businesses on the same billing cadence, and read the cohort retention curve beside the rate.
- What is the difference between customer churn and revenue churn?
- Customer churn counts subscribers lost; revenue churn counts billing lost. They differ because churners are rarely average value. Six hundred subscribers billing $32 leaving a 10,000-strong base that bills $45 on average is 6.00% customer churn and 4.27% revenue churn. Revenue churn exceeding customer churn means the valuable customers are the ones going.
- Why do my churn reports disagree with each other?
- Usually five unstated inputs: which denominator, customers or revenue, monthly or annualised, whether a cancellation is booked when requested or when it takes effect, and whether paused and in-recovery subscribers count in the base. Each choice is defensible alone. Publish the definition beside the number and the disagreement resolves.
In depth: DTC subscription churn benchmarks