Customer Churn: What It Is and How to Reduce It
Customer churn is the loss of existing customers over a defined period. Also called customer attrition, it is the counterpart of customer retention, according to IBM’s definition. The churn rate expresses customer loss as a percentage. Understanding that percentage starts with identifying which customers were eligible to leave and when their relationship ended.
How to calculate customer churn
For a starting-cohort measure, ChartMogul’s standard calculation counts customers lost from the population active at the beginning of the period, excluding customers who joined and left during the period and those who left but reactivated before it ended:
Customer churn rate (%) = Starting customers still lost at period end ÷ Customers active at period start × 100
Use the calculation in this order:
- Choose the reporting period and record the customers active at its start.
- Identify members of that starting population who meet the churn definition at its end.
- Divide the lost count by the starting count.
- Multiply by 100 and report the percentage alongside both counts.
Keep new-customer departures and reactivations visible as separate movements. They help explain customer activity without changing the population measured by this formula.
Count customers consistently. Under ChartMogul’s subscription rules, canceling one subscription while another remains active is a contraction, rather than the loss of the customer.
Define when a customer has churned
For repeat-purchase businesses, a gap between orders can reflect the normal buying cycle. Qualtrics recommends matching the churn definition and observation period to the sales cycle. Set an inactivity threshold using that cycle, document it, and label the result as a measure based on inactivity. An overdue purchase and an explicit cancellation are different observations.
For subscriptions, ChartMogul identifies three recognition dates: the cancellation request, the end of the paid service period, and the subscription’s end in the billing system. These dates can place the same departure in different reporting periods.
Choose the date that fits the report’s purpose. Use a service-end definition to track relationships through their paid term, or a billing-system definition to reconcile with subscription records. Record scheduled cancellations separately when the main rate recognizes churn later.
Document the customer unit, active status, loss event, reporting interval, and treatment of reactivation. Apply the same rules to every period being compared.
Voluntary and involuntary churn
IBM distinguishes voluntary churn from involuntary churn by the mechanism behind the departure.
| Type | Meaning | Reasons to investigate |
|---|---|---|
| Voluntary churn | The customer chooses to stop buying or using the service. | Dissatisfaction, price, a competing offer, or a change in needs. |
| Involuntary churn | The relationship ends because of a disruption beyond the customer’s intended choice. | Payment problems, technical problems, or the service becoming unavailable. |
Classify departures from cancellation records, billing events, and customer feedback. Keep unknown reasons explicit. Investigate payment failures separately from complaints about the product or service.
Customer churn versus revenue churn
Customer churn measures relationships lost; revenue measures show the financial effect. For recurring revenue, Stripe’s retention formulas distinguish gross revenue retention from net revenue retention and exclude revenue from newly acquired customers.
Use these quantities from the same starting customer group:
- R: recurring revenue at the start.
- C: recurring revenue lost through customer departures.
- D: recurring revenue lost through downgrades or reduced purchases.
- E: additional recurring revenue from expansion.
| Metric | Formula | What it measures |
|---|---|---|
| Gross revenue churn | (C + D) ÷ R × 100 | Revenue lost before expansion offsets it. |
| Gross revenue retention (GRR) | (R − C − D) ÷ R × 100 | Starting revenue retained, excluding expansion. |
| Net revenue retention (NRR) | (R − C − D + E) ÷ R × 100 | Revenue retained after expansion and losses. |
Expansion can raise NRR above 100% while customer departures continue, as Stripe explains. A high net retention figure therefore needs to be read alongside customer churn and gross retention.
Specify whether a revenue churn report includes downgrades. Report the revenue effect of retention offers alongside the customer count.
Why customers leave
Qualtrics identifies service quality, product fit, changing needs, pricing, and competing offers as possible churn drivers.
To investigate a departure, record what the customer said, what happened before the loss, and whether a problem remained unresolved. Useful records include cancellation feedback, support conversations, purchase history, product activity, and payment events.
A compact loss record should contain the customer identifier, relationship start, churn date, revenue lost, documented reason, and any attempted recovery. Keep the customer’s stated reason separate from an inferred explanation.
How to analyze churn and identify risk
Cohort analysis groups customers by a common starting point and follows their retention over time. In Amplitude’s retention documentation, the starting event defines a user’s cohort entry date, and a return event defines the behavior being measured.
For customer loss analysis, choose an entry event that matches the relationship, such as the first purchase or paid activation. Compare groups after the same elapsed time, and include only groups that have reached the observation point.
Then separate groups by characteristics relevant to a possible response: relationship age, product, purchase frequency, pricing tier, or recorded departure reason. Keep each group’s starting count and lost count beside its rate.
For digital activity, distinguish returning on a particular day from returning on that day or later (Amplitude treats these as different retention measures). Label activity retention separately from paid-customer retention.
Reduced activity, declining purchases, or repeated complaints can guide investigation; Qualtrics recommends combining operational records with customer experience feedback. Use those signals to identify relationships for review.
An association between a behavior and subsequent departure does not establish that the behavior caused churn. NIST’s statistical guidance makes this distinction. Treat an observed pattern as a candidate explanation to investigate and test.
How to reduce customer churn
Improve the beginning of the relationship
IBM recommends tutorials, FAQs, and self-help guidance to help new customers use products successfully.
Review the steps between purchase and the first useful outcome. Use recorded questions and abandoned steps to choose what to clarify or simplify. Measure whether the revised process improves later retention.
Resolve service and fit problems
Qualtrics recommends customer education and timely support as retention measures.
Prioritize recurring complaints by their frequency, revenue affected, and available remedy. Record what was fixed and check whether the affected customers continue buying. For pricing complaints, distinguish affordability from an unmet need before choosing a response. Assess any discount or smaller plan through both customer retention and revenue retained.
Recover failed payments
For recurring payments, Stripe’s recovery analytics separates first-attempt failures, successful recoveries, and payments still undergoing retries. Recovery methods include retries and customer emails.
Review failure reasons, make payment details easy to update, and provide clear instructions for resolving the failure. Keep an account in the appropriate recovery status until it meets the documented churn rule.
Use complete recovery windows for comparison: Stripe notes that unfinished retry attempts can temporarily lower reported recovery rates. Track recovered payment value separately from customer retention, and follow recovered accounts through the next relevant renewal.
Check whether a retention change worked
Before introducing a change, define the eligible customer group, the outcome, and the observation period. Keep the churn definition consistent throughout the comparison.
Where feasible, assign eligible customers randomly to the existing process or the revised process. NIST’s description of completely randomized designs explains the principle of assigning treatments randomly and comparing response outcomes.
Evaluate departures after the same elapsed time. Include the number of customers observed, revenue retained, and the cost of the intervention. Follow the outcome beyond an accepted offer or recovered payment to assess retention through a subsequent purchase or renewal.
For changes evaluated only against an earlier period, document differences in customer mix, purchase cycles, and concurrent changes. Describe the result as a comparison, with those limitations stated.
What is a good customer churn rate?
A useful benchmark matches the customer population and measurement rules. Recurly recommends comparison with businesses serving similar customer profiles and publishes benchmarks from its own subscription-business network.
Check the reporting interval, customer definition, purchase or billing cycle, and inclusion of voluntary and involuntary losses before using an external rate. The dataset’s population also matters when deciding whether the comparison applies.
Set a retention target for a clearly identified group and period. Pair the target with revenue retained and the cost of the proposed improvement. Review progress against the same definition so a change in counting rules does not become the explanation for a lower rate.