Churn Rate shows what percentage of customers left over a specific period. For a subscription, that means cancellation; for an app, stopping its use; for a service, losing an active customer. Without a period and a clear denominator, churn becomes a pretty number that means nothing.
Updated August 11, 2026.
Customers and money must be counted separately. You can lose few accounts but a lot of revenue if the fattest ones leave. You can lose many small customers and barely notice it in revenue. That is why one Churn Rate is often not enough.
Customer Churn
Formula: customers lost during the period / customers at the start of the period x 100%.
An antivirus product had 1,000 subscribers at the start of the year. During the year, 120 left. Customer churn is 120 / 1,000 x 100% = 12%. Do not quietly mix customers acquired during the period into the denominator. That compares different cohorts and produces garbage.
Gross and Net Revenue Churn
Gross revenue churn measures lost recurring revenue without counting upgrades:
Lost revenue / revenue at the start of the period x 100%.
Net revenue churn also includes downgrades and expansion revenue:
(lost revenue + downgrades – upgrades) / revenue at the start of the period x 100%.
Suppose MRR at the start of the month was 100,000 USD. Churn took 8,000 USD, downgrades took another 2,000 USD, and upgrades added 5,000 USD. Net revenue churn is 5%. If growth from existing customers covers the losses, the metric can turn negative. That is not an error. It was a good month.
How to Compare Churn Across Periods
Five percent a year and five percent a month are catastrophically different. You cannot simply multiply monthly churn by 12. With a steady 5% monthly churn, after one year the remaining share is:
(1 – 0.05)12 = 0.54.
That leaves roughly 54% of the starting cohort, while 46% leaves. You can compare months and years only with the same active-customer definition, the same churn event, and the same cohort logic.
Churn, Retention, and LTV
For the same starting cohort and period, customer retention equals 100% minus customer churn. But that does not make every Retention Rate and churn metric mirror images. Day 7 retention, rolling retention, and calendar churn answer different questions.
The formula LTV = ARPU / churn works only as a shortcut for a stable subscription when ARPU and churn use the same period. Margin, refunds, fees, different plans, and volatile churn break this toy fast. The detailed calculation is in the LTV article.
Why a Media Buyer Needs Churn
For a media buyer, churn shows what happened after the pretty click, lead, or install. If users drop off before their first payment or immediately afterward, a high CR will not save you. The traffic can be real but useless for that monetization model.
Look at churn by source, campaign, GEO, device, creative, and launch cohort. One overall metric for the whole project can easily hide a source sending outright trash.
How to Reduce Churn Rate
- Do not lie in the ad. If the creative promises one thing and the product delivers another, the customer will leave quickly and correctly.
- Get users to the first value. Onboarding should show why the service is useful as quickly as possible.
- Break down cancellation reasons. A failed payment, a bug, poor support, and no value require different fixes.
- Measure cohorts separately. The overall average almost always hides the problem.
- Compare retention cost with value. You do not need to keep an unprofitable customer at any cost.
Reducing churn makes sense as long as retention costs less than the value it preserves. Everything after that is charity.
See the other terms in the CPA glossary.