Subscription churn benchmarks 2026.

Every benchmark on this page states its period, its source, and its date. Where a number is ours, it says which cut of our production data it came from. Where a number is somebody else's, it is linked so you can check it. That sounds like a low bar. It is the bar most churn benchmark pages fail, which is why the answers you find to "is 5% churn good" range from "excellent" to "you are in serious trouble".

Is a 5% churn rate good?

5% per month is high for subscription commerce. 5% per year is excellent. The gap between those two readings is the whole answer.

Churn compounds, so a monthly rate does not multiply by twelve - it decays. Run the arithmetic on a cohort of 1,000 subscribers who never get replaced:

So a store quoting "5% churn" is either losing about a twentieth of its subscriber base every year, or close to half of it. Nobody can tell you which without the period. If you are reading your own dashboard, check what window it uses before you compare yourself to anything.

Against a yearly measure, 5% sits just outside the band Recurly describes as where most well-run subscription businesses operate - 2% to 4% annual churn, with anything above 5% worth investigating regardless of vertical.

Is 20% churn high?

Monthly: yes, severely. 20% a month leaves 69 subscribers out of 1,000 after a year. You would be losing half your subscriber base roughly every three months just to stand still, and no acquisition budget survives that for long.

Annual: no, not unusual. 20% a year is roughly four to five times the 4.25% median annual ecommerce churn Recurly reports across its network, but it is an ordinary number for consumer subscriptions with low price points and impulse signups - which describes most Shopify subscription stores.

If you are seeing 20% and you are not sure which one it is, one check settles it: divide cancellations in a single calendar month by the active subscribers you had on the first of that month. That is a monthly rate. Anything else is a different metric with the same name.

Subscription churn benchmarks by industry

The most complete industry table with a stated period comes from Recurly network data, July 2026. Recurly labels these median annual churn rates, split into voluntary (the subscriber left) and involuntary (the payment failed and never recovered).

Source: Recurly, churn rate benchmarks, median annual churn from Recurly network data, July 2026. Ecommerce is the row that covers subscription boxes and direct-to-consumer.

Two things worth pulling out of that table. First, involuntary churn is about a third of the ecommerce total - 1.38 points of the 4.25. That third is a payments problem, not a product problem, and it is the half you can fix without touching what you sell. Recurly's own network shows involuntary churn collapsing to 0.18% in the highest-value cohort ($250+ average revenue per customer), which is what good retry logic and card-updater coverage buy you.

Second, ecommerce and education sit at the top of the table for the same underlying reason: low price points and discretionary signups. If you sell a $22 monthly box, the SaaS benchmark is not your benchmark.

Why published churn benchmarks disagree by 13x

Recurly puts ecommerce at 4.25% annual. A large number of benchmark pages quote 6.5% to 8.5% monthly for the same category. Compounded, that second range works out to 55% to 66% a year.

Those two claims are 13 to 15 times apart, and both circulate as "the ecommerce subscription churn benchmark". They are not measuring the same thing, and in most cases you cannot tell from the page which one you are reading. Three things go unstated almost every time:

• The period. Monthly and annual rates share a name and differ by an order of magnitude.
• The denominator. Subscribers at the start of the period, average subscribers across it, and contracts rather than customers all give different answers on the same data.
• Whether paused counts as churned. Almost nobody says. It is worth more than you would guess - see the next section.

Before you benchmark yourself against any number, find those three. If a page does not state them, it is not a benchmark, it is a vibe.

The paused subscriber problem, measured

Here is a number we can put a real denominator on. Across 59,964 Shopify subscription contracts running on Joy Subscriptions, this is how contract status is distributed:

Source: Joy Subscriptions production data, aggregated 2026-04-23, anonymised, cohorts of 20+ shops only. Installs observed from 2024-07-24 to 2026-04-23. Published in full in the State of Shopify Subscriptions 2026 report.

15.7% of contracts are paused, not cancelled. That single line decides your churn number:

• Count paused as churned, and the non-active share of this dataset is 63.6%.
• Count paused as retained, and it is 47.7%.

Same contracts, same window, 15.7 percentage points apart, purely from a definition. Any benchmark that does not tell you which convention it used is giving you a range, not a number.

Treating pause as churn is also the wrong read commercially. Recurly's 2026 State of Subscriptions, drawn from 76 million subscribers across 2,200 merchants, reports that three out of four subscribers who pause eventually return, and that merchants offering a pause-before-cancel option saw pause usage rise 337% year over year. A paused subscriber is a retention outcome. Filing them under churn hides the thing that worked.

If you are deciding how to handle this in your own store rather than your spreadsheet, we wrote the subscriber-facing version separately: pause or cancel a subscription.

Voluntary and involuntary churn are different problems

Splitting the two is the highest-value thing you can do to your own churn reporting, and almost no published benchmark does it.

• Voluntary churn is a decision. The subscriber has too much product, hit a cadence mismatch, or stopped seeing value. Recurly's 2026 report finds 52% of consumers cancelled at least one subscription in the past year for lack of use. The fixes are product, onboarding, cadence and save flows.
• Involuntary churn is a failed payment that never recovered. Expired cards, insufficient funds, issuer declines. The fixes are retry schedules tuned per decline code, dunning emails, and card-updater coverage - none of which require changing what you sell. See payment retries and dunning.

On the ecommerce row above, involuntary is 1.38 of 4.25 points - roughly a third of the problem, and the cheaper third to fix.

What this page does not claim

The Joy figures on this page are a status distribution across a roughly 21-month observation window. They are not a monthly churn rate, and we have deliberately not converted them into one. Publishing a cumulative lifetime share as if it were a monthly rate is the single most common error in this category, and a wrong benchmark gets quoted for years.

What is still missing, and what we are working on: monthly cancellations over active contracts at the start of each month, split by industry and by store size, and split voluntary versus involuntary. When that cut is ready it will be published here with the same sourcing discipline as everything above.

How to work out your own number

1. Pick the period and say it out loud. Monthly is the default for subscription commerce. Write "monthly" next to the number every single time.
2. Fix the denominator. Active subscribers on day one of the month. Not the average, not the end-of-month figure.
3. Decide where paused goes, and hold it. Our recommendation: report active, paused and cancelled as three separate states, and never fold pause into churn.
4. Split voluntary from involuntary. If your app cannot tell you which cancellations were failed payments, you cannot tell which half of your churn is fixable this quarter.
5. Compare against your own price point and category, not the cross-industry average. A $22 box and a $250/month B2B subscription do not share a benchmark.

Related reading: churn, customer retention, active subscriptions, and pause subscription.