Written by Krisztián Stark
Key takeaways
- Customer churn rate is the customers lost during a period divided by the customers you had at the start of it, and new sign-ups during the period stay out of both numbers.
- Monthly churn compounds, so annual churn is 1 - (1 - monthly churn) ^ 12: 2% a month is about 21.5% a year, not 24%.
- Treat voluntary and billing churn as separate problems: RevenueCat found billing failures behind nearly a third of Google Play subscription cancellations.
What is churn rate?
Churn rate is the percentage of customers or users who stop using or paying for a product during a set period. The basic formula is customers lost during the period divided by customers at the start of the period, times 100. If an app starts the month with 2,000 subscribers and 140 of them cancel or fail to renew, its monthly churn rate is 7%. Revenue churn applies the same idea to recurring revenue instead of headcount.
Churn is the mirror image of retention: retention tells you who stayed, churn tells you who left and, split properly, why. For the retention side, read What is retention rate?
How to calculate churn rate
The customer churn formula (also called logo churn or subscriber churn) is the one most reports use:
Churn rate = customers lost during the period / customers at the start of the period x 100
Recurly’s churn benchmark page uses the same definition. Three details decide whether your number is right:
- Leave new customers out. People who join during the month are not in the starting count, so track their cancellations separately as first-period churn.
- Pick one definition of “lost”. For subscriptions, a customer is lost when the paid period ends without a renewal, not on the day they press cancel. For free apps, you need an inactivity rule, such as no session for 30 days.
- Keep the period fixed. Monthly and annual churn are different numbers.
Illustrative example: one month of a subscription app
On October 1 an app has 2,000 paying subscribers. During October, 140 of them reach the end of their paid period without renewing, and 300 new people subscribe. Monthly churn is 140 / 2,000 = 7%. The app ends the month with 2,000 - 140 + 300 = 2,160 subscribers, so it grew, but growth hid the churn. Sample figures only.
Monthly vs annual churn
Churn compounds, because each month you lose a share of a base that is already smaller. Recurly makes the same point: a 2% monthly churn rate is roughly 22% a year, not 24%. To convert:
- Monthly to annual: annual churn = 1 - (1 - monthly churn) ^ 12
- Annual to monthly: monthly churn = 1 - (1 - annual churn) ^ (1/12)
| Monthly churn | Simple x 12 (wrong) | Compounded annual churn | Customers left after 12 months, from 1,000 |
|---|---|---|---|
| 2% | 24% | 21.5% | 785 |
| 5% | 60% | 46.0% | 540 |
| 7% | 84% | 58.1% | 419 |
The formula assumes the monthly rate stays the same all year, which it rarely does. Churn is usually highest in the first months, so use it to compare numbers, not to forecast.
Customer churn vs revenue churn
Customer churn counts people. Revenue churn counts money, which matters as soon as customers pay different amounts. Losing 5% of customers who were all on your cheapest plan can cost far less than 5% of revenue; losing one large account can cost far more.
- Gross revenue churn = (recurring revenue lost to cancellations + revenue lost to downgrades) / recurring revenue at the start of the period
- Net revenue retention (NRR) = (starting recurring revenue - cancellations - downgrades + upgrades) / starting recurring revenue
- Net revenue churn = 100% - NRR. It goes negative when upgrades from existing customers outweigh what you lose.
Count only revenue from customers who were there at the start; new customers’ revenue belongs in growth, not in NRR.
Illustrative example: $10,000 in monthly recurring revenue
You start the month at $10,000 of monthly recurring revenue. Cancellations remove $600, downgrades remove $200, and upgrades from existing customers add $500. Gross revenue churn is ($600 + $200) / $10,000 = 8%. Net revenue retention is ($10,000 - $600 - $200 + $500) / $10,000 = 97%, so net revenue churn is 3%. Sample figures only.
If you sell in-app purchases rather than subscriptions, there is no recurring revenue to measure, so track paying-user churn (payers last month who did not pay this month) alongside ARPU.
Types of churn compared
“Churn” covers several different measurements. Name the one you mean in every report.
| Type | What it measures | Formula | Where it matters |
|---|---|---|---|
| Customer (logo) churn | Share of paying customers lost | Customers lost / customers at start | Subscriptions, SaaS, memberships |
| Gross revenue churn | Share of recurring revenue lost, before upgrades | (Canceled + downgraded revenue) / starting revenue | Products with several price tiers |
| Net revenue churn (or NRR) | Revenue change from existing customers, including upgrades | 100% - NRR | Products where customers can expand |
| Voluntary churn | Customers who chose to cancel | Cancellations / customers at start | Product value, pricing, competition |
| Involuntary (billing) churn | Customers lost to failed payments | Billing-failure losses / customers at start | Card and store billing, recovery flows |
| Trial churn | Trials that end without converting | Trials canceled or expired / trials started | Apps with free trials |
| User (activity) churn | Active users who stop using the app | Active in period 1, inactive in period 2 / active in period 1 | Free apps, games, communities |
Voluntary and involuntary churn add up to total customer churn, but the fixes differ: voluntary churn is a product and pricing problem, involuntary churn a payments problem.
Churn rate vs retention rate
For the same group of people and the same period, churn rate = 100% - retention rate: if 85% of your October 1 subscribers still pay on October 31, October churn is 15%. The trouble starts when the numbers come from different groups or definitions:
- Cohort retention is not period churn. D30 retention follows one group of new users from install day. Monthly churn follows your whole customer base, most of whom joined long ago, so the two can tell different stories.
- Classic retention overstates permanent churn. Classic D30 retention counts users active on exactly day 30. Some users missing that day return on day 34, so its complement means “not active on day 30”, not “gone for good”.
- Subscription retention is cumulative. RevenueCat’s report counts a subscription as retained after 12 months only if it has enough paid renewals to cover the time, such as 12 monthly renewals. Its complement is cumulative churn, not a monthly rate.
Stickiness ratios such as DAU/MAU sit next to both: they tell you how often the people who stay actually show up.
Churn in mobile apps
New-user churn: D1, D7 and D30 cohorts
Most app churn happens in the first days after install. Adjust’s widely cited all-category medians are 26% retention on day 1 and 7% on day 30, which means 93% of a typical install cohort is not active on day 30. Our day 1, 7 and 30 retention benchmarks collect those reports, including faster-falling game data.
Active-user churn in free apps and games
For an app without subscriptions, decide what “churned” means for your usage pattern. A daily puzzle game might call a user churned after 7 days without a session; a travel app might wait 90. Then measure monthly: of the users active in September, what share had no session in October? Keep the rule fixed, or your trend will move for reasons unrelated to users.
Subscription churn: trials, renewals and billing
Subscription apps churn at specific moments: at the end of a trial, at each renewal, and whenever a payment fails. RevenueCat’s State of Subscription Apps 2026, built on more than 115,000 apps and covering 2025, shows how front-loaded this is. For 3-day trials, 55.4% of trial cancellations happen on day 0, the day the trial starts. For annual plans, month 1 accounts for 35% of all cancellations in the first year, with another bump in month 12 before renewal.
The same report splits cancellations by cause. Billing errors were behind nearly a third of Google Play subscription cancellations (31%), against 14% on the App Store. On Android especially, a large part of your churn may have nothing to do with how people feel about your app.
What is a good churn rate?
A good churn rate is lower than last quarter’s for the same kind of customer and plan, and low enough that lifetime value covers what you pay to acquire customers. Published benchmarks show whether you are in a normal range, if you compare like with like. Sources checked on October 20, 2026:
| Source | What it measures | Headline figures |
|---|---|---|
| Recurly churn rate benchmarks (July 2026 data) | Subscribers lost across Recurly’s network of subscription businesses | 3.60% overall: 2.34% voluntary, 1.25% involuntary. From 3.22% (SaaS) to 4.99% (education) |
| RevenueCat State of Subscription Apps 2026 (2025 data) | Share of app subscriptions still active after one year, by plan length; category medians | Yearly plans 20% to 40%, monthly 6% to 14%, weekly 1% to 2% |
| Adjust, via our retention benchmarks | New installs still active on day N, all categories | 26% on day 1, 13% on day 7, 7% on day 30 |
Recurly labels its figures median annual churn rates, while its formula example is monthly, so check the period on the page before comparing. The split holds either way: about a third of Recurly’s churn came from failed payments, and involuntary churn fell from 1.30% for customers paying $10 to $25 a month to 0.18% for those paying over $250. RevenueCat’s numbers are survival rates for app store subscriptions: if 8% of monthly subscriptions are still active after a year, about 92% churned at some point in that year.
Plan length changes everything. In RevenueCat’s data, the median first monthly renewal sits in the 53% to 61% range, and the second and third renewals run higher, so the steepest churn sits in the first billing cycles. If your churn looks bad, first check whether your plan mix explains it. CAC vs CPA vs LTV shows how churn feeds into lifetime value, and the LTV calculator lets you try different churn rates against your own revenue per user.
Why users churn and how to reduce it
Onboarding that reaches value fast
Day 0 trial cancellations and weak D1 retention mean people did not reach the moment where the app proved itself. Cut steps before it, delay permission prompts and account walls, and make sure ads promise what the first session delivers.
A reason to come back, and reminders users want
D7 and D30 depend on the core loop: progress, fresh content, social ties, or a weekly job the app does. Find what retained users did in week one that churned users did not, and steer new users toward it. A reminder about something the user cares about (a streak, a saved project, a reply) brings people back; a generic daily nudge gets notifications turned off. Ask for permission after the user has seen value. Rewarded loops such as daily check-ins and levels can support this; rewards app retention covers how to build them without inviting abuse.
Pricing and plan design
Plan length shapes churn more than almost anything else. RevenueCat found yearly plans keep far more subscribers after a year, but annual buyers cancel heavily in month 1. Recurly’s churn page, citing its 2026 State of Subscriptions report, says 38% of consumers prefer pausing over canceling, so offer a pause option.
Billing retries and grace periods
Involuntary churn is the cheapest churn to fix. Both app stores give you tools, checked on October 20, 2026:
- App Store: Apple’s Billing Grace Period lets subscribers keep paid access while Apple keeps trying to collect payment. You can choose 3, 16, or 28 days (weekly subscriptions are capped at 6 days).
- Google Play: Google’s subscription lifecycle documentation says auto-renewing base plans have a grace period and account hold enabled by default, with a default recovery period of 60 days in total. Users keep access during the grace period; during account hold you should block access.
- Your own billing: retry failed charges on a schedule and send a message with a link to update the card.
Win-back after cancellation
Some churn is temporary. RevenueCat found that within a year, 20% of churned monthly subscribers reactivated, against 9% for weekly and 5% for annual plans, and that gaming churners rarely return. Apple’s win-back offers let you target lapsed subscribers by how long they paid and how long ago they left (1 to 24 months), with a free, pay as you go, or pay up front offer. Keep the account and saved progress so returning is easy.
Churn fixes that backfire
Hiding the cancel button, padding the cancel flow, charging after a trial without a clear reminder, or deleting progress the moment a subscription lapses. These raise refunds, chargebacks, and complaints, and can break store or consumer protection rules.
Can rewards reduce churn without bribing users?
Yes, if the reward points users toward the value of your product instead of paying them to stay. A small bonus that brings someone back to finish a lesson supports the habit you want. Paying people not to cancel, or rewarding store ratings, does not: Apple and Google both forbid incentivized ratings and reviews, as the app store policy section of our mobile game guide explains.
In an app with a premium currency, many users never pay, and some leave when they run out of currency. An offerwall lets them earn your currency by completing advertisers’ offers, while you earn per completed offer, so they keep progressing without a purchase. What is an offerwall? explains the model. Keep it within these limits:
- Keep the wall optional. Never make offers a condition for using the app, claiming a bonus, or keeping an account.
- Price the currency so offers earn about what a purchase would buy; How to set your virtual currency exchange rate shows the method.
- Fund bonuses from your own earnings and measure cost per retained user.
- Test it: compare D30 retention and paying-user churn for cohorts with and without access.
Giving non-paying users a way to keep progressing
A Sharklio offerwall lets your users earn your own currency, with your currency name and icon, your conversion rate and user split, and an optional time-limited bonus for re-engagement events. It runs as a signed link or iframe, through the Offers API, or in apps through the Android and iOS SDKs and the Unity, Flutter, and React Native plugins. Every credit and every reversal reaches your server through a signed postback, so balances stay accurate, and your reports show clicks, conversions, and reversals by country, offer, and sub ID. Every publisher application and app is reviewed by hand before the wall goes live. See how the wall works, or get in touch with questions.
Frequently asked questions
How do you calculate churn rate?
Divide the number of customers lost during a period by the number of customers at the start of that period, then multiply by 100. Leave customers who joined during the period out of both numbers.
What is the difference between customer churn and revenue churn?
Customer churn counts the share of customers who leave. Revenue churn counts the share of recurring revenue lost to cancellations and downgrades, so it weighs each customer by what they paid.
How do you convert monthly churn to annual churn?
Use annual churn = 1 - (1 - monthly churn) ^ 12. A 2% monthly churn rate is about 21.5% a year, and 5% a month is about 46% a year.
Is churn rate the same as 100% minus retention rate?
Only for the same group of people over the same period. Cohort retention such as D30 and monthly churn of your whole customer base measure different groups, so their complements do not match.
What is involuntary churn?
Involuntary churn is when a subscriber loses access because a payment fails, not because they chose to cancel. Retries, grace periods, and reminders to update payment details recover a share of it.
What is a good churn rate for a subscription app?
It depends heavily on plan length and category. RevenueCat’s 2026 report found median one-year retention of 20% to 40% for yearly plans and 6% to 14% for monthly plans, so compare yourself with apps on the same plan type and with your own past cohorts.
