All segments

Subscriber Churn Rate: How to Measure It Honestly

Subscriber churn rate is easy to miscalculate: the denominator, gross vs net revenue churn, and involuntary churn all bend the number your dashboard shows.

  • Published
  • Reading time 13 min read
  • Author Nafiul Hasan
Subscriber Churn Rate: How to Measure It Honestly. Diagram: two records, drifting. RETAIN Subscriber Churn Rate: How toMeasure It Honestly SYSTEM ASYSTEM B pointerflow.com

Short answer

Subscriber churn rate is only honest once you fix the denominator (which subscribers count as active), separate gross from net revenue churn, and stop letting involuntary payment failures hide inside a single voluntary-looking headline number. A cohort view then shows what one monthly percentage can't: where in the subscription lifecycle the loss actually happens.

What does subscriber churn rate actually measure?

Subscriber churn rate is the share of subscribers who stopped paying in a given period, expressed against the subscriber count you’re measuring it against — and that second half of the sentence is where most published numbers quietly diverge from each other. Two subscription businesses can report the same 6% churn rate on genuinely different underlying retention, because one counted its denominator at the start of the month and the other averaged the start and end, one included paused accounts as active and the other didn’t, and one measured cancelled logos while the other measured cancelled revenue. None of that is dishonest on its own. It becomes dishonest when the method isn’t disclosed next to the number.

This is written for subscription operators running a DTC consumables brand, a software product, or a membership programme at $3M–$30M in revenue on Recharge, a comparable subscription platform, or a custom billing stack — businesses with enough subscriber volume that the denominator choice moves the reported number by a meaningful margin. A brand under the $3M floor with a few hundred subscribers doesn’t have this problem yet; at low volume, a single cancelled account swings the percentage more than any methodology choice, and the fix there is more subscribers, not a better formula.

How do I pick the right denominator for subscriber churn rate?

“Subscribers who cancelled divided by subscribers you had” sounds like one number, but “subscribers you had” can mean at least four different counts, and each produces a different answer from identical underlying cancellations. Start-of-period count uses subscribers active on day one of the month. End-of-period count uses subscribers active on the last day, which already excludes anyone who joined and left inside the same month. Average of start and end smooths both. Distinct-active-any-day count includes everyone who was a subscriber for even part of the month, the largest and most forgiving denominator of the four.

The practical effect: a business with strong month-over-month growth reports a lower churn rate on a start-of-period denominator than on an average denominator, purely because the start-of-period count is smaller and hasn’t yet absorbed that month’s new signups. A business that’s flat or shrinking sees less distortion from the choice, because the denominator options converge when the subscriber count isn’t moving much.

A benchmark page is only comparable to your own number if it discloses its denominator convention alongside the figure. Check that before treating Pointerflow’s DTC consumables churn benchmark, or any other published number, as directly comparable to your own dashboard.

Gross and net revenue churn drifting apart Two lines starting at the same point. Gross revenue churn rises steadily across the months shown. Net revenue churn stays low because expansion revenue offsets some of the loss, hiding the gross figure underneath it. gross churn net churn

What’s the difference between gross and net revenue churn?

Gross revenue churn is cancelled and downgraded revenue as a share of starting revenue, full stop — it never gets offset by anything. Net revenue churn subtracts expansion revenue (upgrades, added seats, cross-sold subscription add-ons) from that same cancelled-and-downgraded figure before dividing by starting revenue, which means net revenue churn can go negative: a subscription business can lose accounts and still show negative churn if the accounts that stayed spent enough more to cover the loss.

Both numbers are honest. They answer different questions. Gross revenue churn tells you how much retention work the product and support team have to do regardless of what sales or merchandising achieves elsewhere; net revenue churn tells you whether the whole subscription programme is growing or shrinking, expansion included. The mistake is reporting only net, because a healthy net number can sit directly on top of a genuinely deteriorating gross number if expansion revenue happens to be covering for it that quarter — and expansion revenue concentrated in a handful of large accounts is a fragile thing to depend on.

A worked distinction, illustrative rather than measured: say starting MRR for a cohort is $100,000, cancelled and downgraded MRR that month is $6,000, and expansion MRR from existing accounts upgrading is $4,500. Gross revenue churn is $6,000 ÷ $100,000, or 6%. Net revenue churn is ($6,000 − $4,500) ÷ $100,000, or 1.5%. Both describe the same month truthfully; only one of them tells the retention team it has a 6% problem to fix.

How does involuntary churn hide inside your headline number?

Involuntary churn is a subscriber who didn’t choose to leave — a card expired, a bank declined a routine charge, an issuer’s fraud filter flagged a recurring merchant — as opposed to voluntary churn, where someone actively cancelled. Most subscription platforms and most churn dashboards bucket both into the same “cancelled” or “churned” count by default, because from the billing system’s point of view a lapsed subscription looks identical whether the customer clicked cancel or their card simply failed and was never retried successfully.

That matters because the two have completely different fixes. Voluntary churn is a product, pricing or fit problem — the fix is the offer, the product, or a win-back message that addresses why someone actually left. Involuntary churn is a payments and dunning problem — the fix is retry logic, updated-card prompts and expiry-date reminders, none of which touch the product at all. Report one blended churn number and a team can spend a quarter running win-back campaigns aimed at people who never chose to leave, while the real fix, a better retry schedule, sits untouched in the billing settings.

Separating the two doesn’t require a new figure so much as a flag: tag each cancellation event with its cause at the point the billing platform or payment processor reports it — an active cancellation request versus a failed-payment lapse after retries are exhausted — and report the split alongside the blended number rather than instead of it. A blended churn rate that quietly contains a meaningful involuntary component looks like a satisfaction problem when a real share of it is a retry-schedule setting.

Why does a single monthly number hide what a cohort view shows?

A monthly churn rate is a snapshot: cancellations that happened this month divided by subscribers you had. It can’t tell you whether those cancellations came from subscribers in their first month, their sixth, or their second year, because the calculation deliberately collapses every subscriber’s tenure into one shared denominator. Two businesses can report an identical 6% monthly churn rate while having completely different underlying retention shapes — one losing subscribers steadily across the whole lifecycle, the other losing a large share in month one and then holding the rest almost indefinitely.

The difference matters for what happens next. A business with steady, flat monthly churn across tenure has a general retention problem that a lifecycle campaign might not fix, because the loss isn’t concentrated anywhere specific. A business with a front-loaded churn shape has a much narrower, more fixable problem — usually onboarding, first-box fit, or a mismatch between the acquisition offer and the product — and a cohort view is the only calculation that shows which one you actually have.

How do you build a cohort view of subscriber churn rate?

Building a cohort view means tracking each signup month as its own group and following what share of that group is still subscribed at each subsequent month, rather than pooling every subscriber together into one monthly percentage. First, group subscribers by their first-charge month, not their signup or trial-start date, so month zero means the same thing for every cohort. Second, for each cohort, record the count still active at month one, month two and onward — active meaning a charge was successfully collected, not merely that the subscription status field hasn’t been changed to cancelled. Third, express each cohort-month as a percentage of that cohort’s original size, which gives a retention curve rather than a single rate. Fourth, lay cohorts side by side by signup month to see whether the curve shape is improving, worsening, or flat as onboarding, pricing or box contents change over time.

Working through the denominator and cohort choices by hand is exactly what Pointerflow’s subscription churn calculator is for — it walks through the denominator and churn-type decisions before returning a number, rather than assuming one convention silently the way a single-cell spreadsheet formula does.

What breaks in the calculation at volume — multiple plans, pauses, reactivations?

Pause-and-resume features common on Recharge and comparable platforms break a simple active/cancelled churn calculation, because a paused subscriber is neither actively billing nor formally cancelled. Decide upfront whether a pause counts as churned for the period it’s paused, counts as still-active, or gets tracked as a third state, and apply that decision consistently, because switching definitions mid-year makes month-over-month comparison meaningless even though nothing about actual retention changed.

Reactivations — a lapsed subscriber who comes back, whether through a win-back email or on their own — create a similar ambiguity: does a reactivation count as a new subscriber for growth purposes, reversed churn for the original cohort, or both? Most defensible practice treats a reactivation as a new subscriber in a new cohort for churn-rate purposes, while noting separately that they’re a returning customer for lifetime-value and marketing-attribution purposes. Mixing the two calculations understates true new-subscriber churn and overstates how many of the month’s “wins” are genuinely new relationships.

Multiple pricing plans compound both issues, because a churn rate blended across a $19 plan and a $79 plan describes neither cohort accurately, the same way a blended LTV number would. Calculate churn per plan tier where volume allows it, and only roll the tiers up into one blended figure for an executive summary, never as the number a retention team works from.

A subscription business migrating billing platforms mid-year, moving from a legacy platform to Recharge or a comparable system, creates its own denominator problem: cancellations recorded under the old platform’s methodology don’t reconcile automatically with the new platform’s active count, so the migration month typically needs a note of its own rather than blending into a single-tool cohort table. Treat the migration month as a boundary, the same way a mid-cohort price change is treated, and confirm with both platforms’ support teams how each one counted a subscriber as active on the cutover date before publishing a churn number that spans it.

What edge cases distort the subscriber churn rate calculation?

Trial-to-paid conversion timing is the first edge case: if trial cancellations sit in the same denominator as paid-subscriber cancellations, a product with a generous trial window shows an inflated churn rate that has nothing to do with paying-customer retention. Exclude trial-only cancellations from the paid churn calculation and report trial-to-paid conversion as its own separate figure.

Free-to-paid downgrades are the second: a subscriber who drops from a paid tier to a free tier hasn’t cancelled in the traditional sense, but has stopped paying, which is the thing revenue churn is supposed to capture. Decide whether the churn definition is logo-based (did the account still exist) or payment-based (did the account still pay); if it’s payment-based, a downgrade to free counts as churn even though the account technically remains active.

Bulk cancellations from a single account holder managing multiple subscriptions — a gifting programme, a corporate account with several seats — are the third: cancelling five seats under one account holder can look like five churn events or one, depending on whether the denominator counts subscriptions or subscribers, and a spike from a single large account can distort a month’s reported rate in a way that looks like a trend but is really one decision by one customer.

Failed and retried payments are the fourth, and the one most likely to distort the headline number without anyone noticing. Baremetrics has reported that roughly 9% of monthly recurring revenue is lost to failed payments across the subscription businesses it measures (vendor-reported), and a subscription that fails, retries and eventually recovers days later can sit in an ambiguous state that a churn calculation either counts as churned too early or misses entirely if it recovers before the reporting cutoff. Confirm with your payments provider exactly when a failed-and-retrying subscription is marked churned in your own dashboard, because that cutoff date is a methodology choice, not a fact.

What do published churn benchmarks like Recurly’s or Recharge’s actually measure?

Published benchmark reports, including Recurly’s Consumer Goods Churn Benchmark and Recharge’s own DTC subscriber panel data, are useful for context and dangerous for direct comparison unless you read past the headline percentage to the methodology behind it. Both are real, published industry reports (vendor-reported); this page doesn’t reproduce their current figures, because a benchmark number is only meaningful next to the report’s own denominator and churn-type definitions, which change between report editions.

What a benchmark headline typically statesWhat to check before comparing your own number to it
A single blended churn percentage for a category or platform cohort (vendor-reported)Whether it’s gross or net revenue churn, or a logo-count churn rate — the three aren’t interchangeable
A reporting period, usually monthly or annual (vendor-reported)Whether the denominator is start-of-period, average, or distinct-active-any-day
A subscriber base definition such as “active subscribers on the platform” (vendor-reported)Whether paused, trial and involuntary-lapsed accounts are included or excluded from that base

The pattern holds across most published benchmarks: the headline number is real and vendor-reported, but it’s only comparable to your own once the denominator and churn-type definition have been matched to the calculation you’re running — exactly the check to make before quoting a benchmark in a board deck as if it were measuring the same thing your dashboard shows.

Where does churn measurement break in practice on a Tuesday?

The failure looks like this: a retention lead pulls last month’s churn number for a Monday standup, and the dashboard shows a jump from 5% to 8% with no explanation. Nobody notices until Wednesday that a batch of card-on-file updates failed after a payment processor’s routine reauthorisation cycle, pushing a wave of involuntary lapses into the same week — a payments issue, not a satisfaction problem, showing up as a headline churn spike that triggers a win-back campaign aimed at the wrong cause entirely.

The second common break: someone changes the subscriber-count denominator, switching a dashboard filter from “active” to “active or paused” while building a new report, and the churn rate shifts by a percentage point or two with no note in the changelog. Three months later, a quarter-over-quarter comparison quietly compares two different denominators without anyone flagging it, and a retention initiative gets credited or blamed for a change that was really a spreadsheet setting.

Who shouldn’t trust a single blended subscriber churn rate?

A subscription business with under six months of cohort history, fewer than a few hundred active subscribers, or a mix of pricing tiers and pause states that haven’t been split out yet shouldn’t treat one monthly percentage as decision-grade. At low volume, one or two cancellations move the rate enough that the number is closer to noise than signal, and the fix is time and volume, not a different formula. This is written for the $3M-plus subscription operator with enough subscriber volume and history for the denominator and cohort choices to actually matter.

Getting a churn number a team can actually act on is a subscription retention problem before it’s a reporting exercise: the denominator convention, the gross-versus-net split, and the involuntary-versus-voluntary tag all need to be fixed once and applied consistently, not re-decided every time someone pulls a dashboard. That consistency is what Pointerflow’s subscription retention service is built to address — separating the payments fix from the product fix instead of routing both problems through the same win-back campaign.

Sources

  • Baremetrics: approximately 9% of monthly recurring revenue lost to failed payments, reported across its subscription-business customer base (vendor-reported).

Frequently asked

How do I pick the right denominator for subscriber churn rate?

Pick a denominator convention — start-of-period, end-of-period, average of the two, or distinct-active-any-day — and disclose it next to every churn figure reported. Start-of-period gives the smallest, most flattering denominator for a growing subscriber base; distinct-active-any-day is the most forgiving. Neither is wrong, but switching between them month to month makes trend comparisons meaningless.

Should I report gross or net revenue churn as my subscriber churn rate?

Report both, but lead with gross revenue churn for internal retention decisions. Gross churn (cancelled and downgraded revenue over starting revenue) shows the retention team's real workload; net revenue churn subtracts expansion revenue and can mask a deteriorating base behind a few large accounts upgrading. A board summary can show net; a retention team should work from gross.

Why does my churn rate look different in every reporting tool?

Reporting tools differ on denominator convention, on whether paused or trial subscribers count as active, and on whether they bucket involuntary payment failures with voluntary cancellations by default. Two tools measuring the identical subscriber base can produce noticeably different percentages purely from these settings, not from any disagreement about who actually cancelled.

How can I tell how much of my churn is involuntary?

Tag each cancellation event with its cause at the point your billing platform or payment processor reports it: an active cancellation request is voluntary, while a lapse after exhausted payment retries is involuntary. Most dashboards don't split this by default, so the split usually has to be built into your reporting rather than pulled ready-made from a subscription platform's summary screen.

What's the difference between logo churn and revenue churn?

Logo churn counts cancelled subscriber accounts as a share of total accounts, treating a $19 plan and a $199 plan cancellation identically. Revenue churn counts cancelled dollars as a share of starting revenue, weighting toward higher-paying accounts. A business can show flat logo churn while revenue churn worsens, if the accounts leaving are disproportionately high-value ones.

Do paused subscriptions count as churned?

Whether a pause counts as churned depends on a rule you set, not a fact about the subscription. Some teams treat a pause as churned for the paused period since no revenue is collected; others track it as a distinct third state, separate from both active and cancelled. Either is defensible if applied consistently across every reporting period.

Should a reactivated subscriber count as reversed churn?

Most defensible practice treats a reactivated subscriber as a new subscriber in a new cohort for churn-rate purposes, while separately noting they're a returning customer for lifetime-value and marketing-attribution reporting. Counting a reactivation as reversed churn in the original cohort understates that cohort's true churn and can make a win-back campaign look like it fixed retention it didn't.

What does a cohort retention curve show that a monthly percentage doesn't?

A monthly churn percentage pools every subscriber's tenure into one number, so it can't show whether losses concentrate in a subscriber's first month or spread evenly across years of tenure. A cohort retention curve tracks each signup month separately across time, showing exactly where in the subscriber lifecycle the loss happens — information a single blended rate discards by design.

How often should subscriber churn rate be recalculated?

Recalculate the headline monthly rate every billing cycle, but rebuild the full cohort view whenever onboarding, pricing or acquisition offers change, since that's when the curve shape is most likely to move. A cohort view reviewed only annually will miss the early signal that a specific change, like a new welcome flow, affected retention.

Can I compare my churn rate directly to a published industry benchmark?

Only if you've confirmed the benchmark uses the same denominator convention and the same churn type — gross or net revenue, or logo count — as your own calculation. A published industry figure and your own dashboard figure can carry the same label and measure genuinely different things, so match the methodology before treating the comparison as meaningful.

What counts as an active subscriber for churn purposes on Recharge or a similar platform?

Check your platform's own definition of an active subscription status, since 'active' can include or exclude paused, past-due or in-dunning accounts depending on the setting. On Recharge and comparable platforms this is usually a configurable filter, not a fixed default, so confirm it explicitly rather than assuming the dashboard's default view matches your churn definition.

What should I check first if churn suddenly spikes in one week?

Check for a payments-side event first: a card network's routine reauthorisation cycle, a processor outage, or an expired-card wave can push a batch of involuntary lapses into a single week and read as a satisfaction problem when it's a payments problem. Confirm the spike's cause before launching a win-back campaign aimed at the wrong fix.

Who shouldn't rely on one blended monthly churn number?

A subscription business with under six months of cohort history or only a few hundred active subscribers shouldn't treat one monthly percentage as decision-grade, because a handful of cancellations moves the rate more than any real retention trend would. Below that volume, the fix is more time and more subscribers, not a different formula.

Does trial cancellation count toward subscriber churn rate?

Trial cancellations shouldn't sit in the same denominator as paid-subscriber cancellations, because a generous trial window inflates the reported rate with people who never became paying customers. Exclude trial-only cancellations from the paid churn calculation and report trial-to-paid conversion as its own separate figure instead of folding it into churn.

Next step

Is this your subscription retention problem, or a symptom of another one?

Bring your numbers — the churn split, the decline rate, whatever your flows are earning — and we will tell you which of them is the expensive one.

Book a call →