Two Kinds of Churn, Two Different Diagnoses
Most founders make the same mistake with churn: they look at one number where they actually need two. Logo Churn and Revenue Churn are not two ways to look at the same thing. They answer different questions and require different action.
- Logo Churn: the percentage of customers lost over a period. It counts logos, not dollars.
- Revenue Churn: the percentage of recurring revenue lost over a period through churn and downgrades. It counts dollars, not logos.
These combinations tell very different stories:
- High Logo Churn, low Revenue Churn: small customers are leaving while larger accounts stay.
- Low Logo Churn, high Revenue Churn: customers are staying but downgrading, so perceived value is eroding.
- Both high: there is likely a systemic product or market problem.
- Both low: the business is healthy.
If you calculate only one, you are looking at an incomplete picture.
Definitions and Core Components
Logo Churn Rate
Logo Churn Rate = Customers lost during the period / Customers at the start of the period × 100
Example: 500 customers at the start of the month, 15 lost during the month, so Logo Churn = 3%.
What counts as a lost customer must be defined clearly and kept stable over time.
Count as churned:
- Subscription canceled and no longer in grace period.
- Payment recovery failed after grace period and all dunning attempts.
- Customer moved to a $0 plan, if freemium is treated as non-paying.
Do not count as churned:
- Paused subscriptions.
- Customers still in grace period.
- Downgrades, which belong to contraction rather than churn.
Grace Period and Churn Timing
Grace period is the time between the first failed payment and final account cancellation. Customers should not be counted as churned until that period has actually expired.
Day 0 failed payment → active in grace period → retries and notices → churn only after grace period expires
If you record churn at the first failed payment, every technical billing issue inflates churn and every later recovery creates fake reactivation.
Revenue Churn
Revenue Churn should include both full churn and contraction.
Revenue Churn = (Churned MRR + Contraction MRR) / MRR(start) × 100
Using only Churned MRR understates real recurring revenue loss. If $6,000 churned and $4,000 contracted on a $200,000 base, the correct Revenue Churn is 5%, not 3%.
Gross Revenue Retention
GRR = (MRR(start) − Churned MRR − Contraction MRR) / MRR(start) × 100
GRR can never exceed 100% because it excludes expansion. That is exactly why it is one of the most honest retention metrics.
Voluntary vs Involuntary Churn
This split is even more important than Logo vs Revenue. Without it, you cannot prioritize the right fix.
Voluntary churn means the customer intentionally canceled. Common causes are poor perceived value, better alternatives, weak onboarding, pricing friction, or changes in the customer's own business.
Involuntary churn means the customer was lost because billing failed: expired card, insufficient funds, bank decline, or processor issue.
Churned MRR = Voluntary Churned MRR + Involuntary Churned MRR
In SMB SaaS, involuntary churn can account for a significant share of total lost MRR. Product fixes do not solve that. Billing recovery does.
Typical dunning sequence:
- Day 0: failed payment and first email.
- Day 1: retry attempt.
- Day 3: retry plus update-card email.
- Day 7: retry plus SMS or push alert.
- Day 14: last attempt, grace period ends, account churns.
Smart retries should depend on the decline code.
Early Churn
Early Churn is churn in the first 30 to 90 days after signup. It is a separate diagnostic category.
Early Churn Rate (30d) = Customers lost in the first 30 days / All new customers in the period × 100
High Early Churn with otherwise normal Logo Churn usually means the product is fine for customers who understand it, but onboarding is failing to get new users to value quickly enough.
Annual Churn for Annual Contracts
For annual contracts, monthly Logo Churn can look artificially low because customers cannot leave before renewal. The real churn event happens at renewal time.
Annual Renewal Rate = Renewed contracts in period / Expired contracts in period × 100
Annual Logo Churn = 1 − Annual Renewal Rate
If 80 annual contracts expire in December and only 68 renew, Annual Renewal Rate is 85% and Annual Logo Churn is 15%.
Derived Metrics
Customer Lifetime
Customer Lifetime (months) = 1 / Monthly Logo Churn Rate
This is only a rough approximation because real churn curves are not flat.
LTV
LTV = ARPA × Gross Margin / Monthly Churn Rate
Churn has a nonlinear effect on LTV. Reducing churn from 5% to 3% can increase LTV dramatically.
Net Revenue Churn
Net Revenue Churn = (Churned MRR + Contraction MRR − Expansion MRR) / MRR(start) × 100
If expansion exceeds churn plus contraction, Net Revenue Churn turns negative. That is net negative churn.
Churn Cohort Analysis
Cohort Churn Rate(month t) = 1 − Retention Rate(month t)
Churn is not constant over time. Cohort analysis shows the true shape of churn, especially the higher loss rates in the earliest months.
Weighted Logo Churn
Weighted Logo Churn = Σ(MRR of churned customers) / Σ(MRR of all customers) × 100
This is effectively Revenue Churn without contraction. It helps expose customer-size skew behind a headline logo number.
Churn by Segment
Aggregate churn often hides opposite realities inside different segments. Enterprise can look healthy while SMB is deteriorating fast. The same applies to Revenue Churn.
Common Churn Calculation Mistakes
- Including trials in the denominator. Trial users are not paying customers.
- Failing to separate churn and contraction. A downgrade is contraction, not churn plus new revenue.
- Recording churn at first payment failure. Customers remain active during grace period.
- Ignoring seasonality. Month-over-month churn can be distorted by budget cycles and contract timing.
- Reading raw Logo Churn without customer-size context. A $50 account and a $5,000 account are not equivalent losses.
- Rewriting historical churn after reactivation. A return next month does not change the fact that churn happened this month.
Worked Example with Diagnosis
Start of month: 800 customers, $160,000 MRR
| Event | Count | Logo | MRR |
|---|---|---|---|
| New customers | 40 | +40 | +$8,000 |
| Voluntary cancellations | 18 | −18 | −$4,500 |
| Involuntary churn after failed dunning | 8 | −8 | −$1,200 |
| Downgrades (customers stayed) | 12 | 0 | −$2,400 |
| Reactivation | 3 | +3 | +$600 |
| Upsells | 15 | 0 | +$3,000 |
Logo Churn Rate: (18 + 8) / 800 × 100 = 3.25%
Voluntary Logo Churn: 2.25%
Involuntary Logo Churn: 1.00%
Churned MRR: $5,700
Contraction MRR: $2,400
Expansion MRR: $3,000
Revenue Churn: (5,700 + 2,400) / 160,000 × 100 = 5.06%
GRR: 94.94%
Net Revenue Churn: (5,700 + 2,400 − 3,000) / 160,000 × 100 = 3.19%
NRR: 96.81%
MRR(end): $163,500
Diagnosis: headline Logo Churn is above normal, but nearly one-third of it is involuntary. That means a meaningful part of the problem is billing recovery, not product. Revenue Churn being higher than Logo Churn also suggests that larger-than-average customers are being lost or downgraded.
How to Reduce Churn
For voluntary churn:
- Improve onboarding and reduce time to value.
- Use early warning signals from declining product usage.
- Collect exit-survey data at cancellation.
- Offer pause instead of cancellation where appropriate.
- Offer downgrade paths before a full exit.
For involuntary churn:
- Use smart dunning based on decline reason.
- Notify customers 7-14 days before card expiration.
- Include a payment-update link in every billing email.
- Use SMS or push for critical payment failures.
- Enable card updater and adjust grace period by payment history where possible.
Dnoise automatically separates Logo Churn into voluntary and involuntary based on Stripe cancellation and recovery signals. Contraction is tracked separately from Churned MRR, grace period is configurable, and Early Churn is exposed through 30, 60, and 90 day retention views.
Revenue Churn and GRR update in real time with segmentation by customer group.
CTA
Dnoise calculates churn from source data and separates the churn types that require different action.