Metrics & Analytics

Churn Rate

The percentage of customers or revenue lost in a given period — the primary indicator of whether a product retains the value it delivers.

What is Churn Rate?

Churn rate measures the percentage of customers (or revenue) that stops using a product in a given period. It is the inverse of retention and the most direct signal of whether your product delivers lasting value.


Types of churn

TypeFormulaWhat it measures
User churn(Users lost ÷ Users at start) × 100% of users who stopped using the product
Revenue churn (MRR churn)(MRR lost ÷ MRR at start) × 100% of recurring revenue lost
Net revenue churn(MRR lost − expansion MRR) ÷ MRR at startChurn net of upsells (can be negative)

Benchmarks (B2B SaaS)

StageMonthly churn benchmark
Early stage (< $1M ARR)< 5% monthly
Growth stage ($1–10M ARR)< 2% monthly
Scale stage (> $10M ARR)< 1% monthly

Annual churn = approximately (1 − (1 − monthly churn)^12) × 100


Root causes of churn

  • Poor onboarding — users never reach activation
  • Product doesn't solve the stated problem
  • Better alternative exists (or customer perceives it does)
  • Pricing misalignment — expected more for the cost
  • Change in customer's internal priorities or budget

Frequently asked questions

What's the difference between churn rate and retention rate?

They're complements: if monthly churn is 3%, monthly retention is 97%. Retention measures who stayed; churn measures who left. SaaS investors typically focus on retention cohorts (Day 30, Day 90 curves) rather than monthly churn because cohort curves reveal the depth of engagement.

What's negative net revenue churn?

Negative net revenue churn means your existing customers expanded (upgraded, added seats) more than the revenue lost from churned customers. It's a strong signal of PMF — your installed base is growing without new customer acquisition. Stripe and Datadog famously achieved this early.

Apply Churn Rate to your real product data

PMRead ingests customer feedback, interviews, and Slack threads — and generates PRDs grounded in real evidence.

Try PMRead free →