Customer Churn & Revenue Retention
Table of Contents
Customer churn measures how many customers a business loses, but it does not tell you how many revenue dollars disappear. Losing one large account can matter more financially than losing several small ones. Meanwhile, upgrades, price increases and new customers can keep headline growth strong even as part of the existing customer base erodes.
For investors and business owners, the useful question is how much growth comes from durable customer relationships and how much requires replacing lost business. This guide separates customer retention, gross revenue retention and net revenue retention, reconciles a hypothetical subscription cohort, and explains why an attractive LTV/CAC ratio still depends on strong assumptions. These operating measures complement fundamental analysis; they do not replace financial statements or a valuation.
What Do Customer Churn and Retention Measure?
Customer churn, also called logo churn in business subscriptions, counts lost customer relationships. Revenue churn measures lost revenue from those relationships and, under a common gross-churn convention, downgrades by customers who stay. Net revenue retention also includes growth from the opening customers. New customers do not belong in that opening-cohort retention calculation.
A cohort is a defined group held fixed for measurement. It might contain all paying customers at the start of a month, or customers acquired in one particular quarter. Those are different cohorts. An acquisition-cohort table follows customers of similar tenure; an opening-base table measures the entire existing base over a selected interval.
This framework works most directly when a customer can observably cancel or fail to renew a contract. At a retailer, a customer who makes no purchase this month might simply be between purchases. A repeat-purchase rate is therefore different from observed contractual survival. [5]
Set the Measurement Rules First
Before calculating a percentage, establish these conventions:
- Customer unit: one organization, account, household, subscriber or contract? One customer cancelling a single product may still retain another paid product.
- Opening cohort and window: identify the opening customers and exact closing date. Keep new customers outside that cohort.
- Revenue basis: use consistently defined monthly recurring revenue (MRR), annual recurring revenue or recognized revenue. MRR is a monthly-normalized recurring run rate; it is not automatically the month’s accounting revenue or cash collections.
- Movement rules: distinguish full customer loss, contraction among customers who stay, and expansion. State whether you count intermediate cancellation events or compare opening and closing states.
- Returns and boundary changes: disclose reactivation, acquisitions, customer-account consolidation, free trials, delinquent accounts and foreign-currency treatment.
The SEC’s guidance on company metrics emphasizes definitions, calculation methods and context, including relevant assumptions and material methodology changes. A familiar label alone does not make two companies’ numbers comparable. [3]
Customer Churn and Revenue Retention Formulas
For the fixed opening-customer convention used here, let N0 be opening paying customers and L the number of those customers with no paid relationship at closing. Count each customer once:
Customer retention = (N0 − L) / N0 × 100%
For revenue retention, define four amounts on the same basis:
- B: opening recurring revenue from the opening customer cohort.
- C: opening recurring revenue lost when opening customers leave completely.
- D: recurring-revenue contraction among opening customers who remain.
- E: recurring-revenue expansion among opening customers who remain.
C, D and E are nonnegative amounts. Here each customer’s opening-to-closing change is assigned once: a lost customer’s entire opening amount goes into C, rather than also being counted as contraction. For a surviving customer, use the net increase or decrease. Stripe’s revenue-churn explanation similarly distinguishes losses, including downgrades, from expansion that offsets those losses. [6]
Under these conventions, gross revenue churn = (C + D) / B and net revenue churn = (C + D − E) / B, expressed as percentages. They equal 100% minus GRR and NRR, respectively. Net churn can be negative when expansion exceeds cancellations and contraction. Neither revenue-retention formula is defined when B is zero.
Customer Retention vs Gross and Net Revenue Retention
| Measure | What it weights | What it reveals |
|---|---|---|
| Customer retention | Each defined customer equally | How many opening relationships remain |
| Gross revenue retention | Opening revenue, with expansion excluded | How much of the opening revenue base avoids cancellation and contraction |
| Net revenue retention | Revenue from the opening cohort, including expansion | Whether that cohort’s revenue grows or shrinks overall |
Under the formulas above, GRR cannot exceed 100%, while NRR can. A high NRR alongside low GRR means expansion offsets substantial leakage. Strong dollar retention alongside weak customer retention can also indicate reliance on a smaller number of larger accounts. Check customer concentration before treating that combination as broad-based loyalty.
There is no universal good churn rate. The useful comparison matches contract length, customer size, product, geography, pricing model, cohort age and calculation method. A monthly figure and an annual figure are different horizons. Multiplying monthly churn by 12 does not produce a cohort’s annual loss rate; even the compounded expression requires assumptions about renewal opportunities and future churn.
A Customer Cohort Example That Reconciles
Consider a fictional subscription business with 10 paying customer organizations at the opening of November 1, 2026. Their combined MRR is $100,000. Compare those same customers at the close of November 30. All figures use constant U.S. dollars and fixed recurring plans; exclude one-time charges, taxes, acquisitions and currency movements.
This is an endpoint comparison. An opening customer that temporarily cancels and returns before closing stays in the opening cohort at its closing MRR. A formerly inactive customer absent at opening is treated separately as a reactivation. All four rows below concern only the opening customers.
| Opening customers’ outcome | Customers | Opening MRR | Closing MRR |
|---|---|---|---|
| Expanded | 3 | 35 | 65 |
| Unchanged | 4 | 30 | 30 |
| Contracted but still active | 2 | 15 | 10 |
| Lost completely | 1 | 20 | 0 |
| Opening cohort total | 10 | 100 | 105 |
Nine of the 10 opening customers remain. Customer churn is 1 / 10 = 10%, and customer retention is 9 / 10 = 90%. The lost customer supplied $20,000 of opening MRR, so cancellation-only revenue churn is $20,000 / $100,000 = 20%. Equal customer counts do not imply equal revenue weights.
Contraction: D = $15,000 − $10,000 = $5,000.
Expansion: E = $65,000 − $35,000 = $30,000.
GRR: ($100,000 − $20,000 − $5,000) / $100,000 = 75%.
NRR: ($100,000 − $20,000 − $5,000 + $30,000) / $100,000 = 105%.
Gross revenue churn: ($20,000 + $5,000) / $100,000 = 25%.
Net revenue churn: ($20,000 + $5,000 − $30,000) / $100,000 = −5%.
The opening cohort grows by $5,000 of MRR overall, but only after expansion offsets $25,000 of cancellations and contraction. NRR of 105% does not mean every customer stayed, every account expanded or gross retention was strong.
How Headline Growth Can Hide Customer Losses
Continue the same hypothetical month. Four genuinely new customers contribute $40,000 of closing MRR, and one previously inactive customer, absent from the opening base, returns with $5,000. Neither belongs in the opening-cohort retention numerator. Reconcile the entire business separately:
| Bridge item | MRR (USD thousands per month) |
|---|---|
| Opening business MRR | 100 |
| Lost opening customers | −20 |
| Contraction | −5 |
| Expansion | +30 |
| New customers | +40 |
| Previously inactive customers reactivated | +5 |
| Closing business MRR | 150 |
The arithmetic is $100,000 − $20,000 − $5,000 + $30,000 + $40,000 + $5,000 = $150,000. Total MRR growth is ($150,000 − $100,000) / $100,000 = 50%. Closing paying customers number 10 − 1 + 4 + 1 = 14, even though one opening customer was lost.
Thus 50% MRR growth coexists with 75% GRR and 90% customer retention. Acquisition and expansion can obscure the need to replace lost business. This single month demonstrates leakage; calling retention deteriorating requires comparable earlier periods. It also shows run-rate growth, not 50% growth in recognized November revenue. Use the income statement to check actual reported revenue and profit margins.
The financial follow-up is whether acquiring, serving and expanding those accounts produces enough cash after costs. Discounted onboarding, costly renewals or low-margin expansion could make the same MRR bridge less attractive. Check free cash flow rather than treating retained dollars as profit.
Real Company Retention Definitions: Workiva and Snowflake
Issuer disclosures show why the simplified formulas cannot be applied mechanically to every reported metric.
Workiva: Read What “Gross” Includes
Workiva reported gross retention of 97.4% and net retention of 109.6% as of September 30, 2026. Its gross calculation weights surviving prior-year base customers using their prior-year-quarter subscription and support revenue. The denominator uses that prior-year revenue for all base customers. It therefore does not directly measure surviving customers’ current-period downgrades in the way our contraction-adjusted GRR does. [2]
Workiva’s net calculation instead uses current-quarter revenue from surviving base customers and includes add-ons and pricing changes. Its reported net retention declined from 113.6% a year earlier to 109.6%. Read those figures under Workiva’s definitions; do not interpret the 97.4% gross figure as proof that cancellations plus downgrades cost only 2.6%. [2]
Snowflake: Trailing Revenue, Not Monthly Logo Survival
Snowflake reported 126% net revenue retention for the quarter ended July 31, 2026. It defines a trailing two-year window, selects capacity-contract customers using the platform in the first month of year one, and divides that cohort’s second-year product revenue by its first-year product revenue. Customers with no usage in year two remain in the calculation at zero revenue; company events can trigger adjustments. [1]
That is a consumption-revenue comparison across two years, not a single-month MRR comparison or a count of customers that survived. The 126% figure alone does not disclose logo churn. Ranking it against Workiva’s retention, or against the fictional 105% monthly NRR, would mix different measurement designs.
What Churn Can—and Cannot—Tell You About LTV/CAC
Customer lifetime value (LTV) estimates the value of future customer cash contributions. Customer acquisition cost (CAC) measures acquisition spending per acquired customer under a stated cost-allocation method. For an LTV/CAC ratio, specify whether LTV is before acquisition cost; subtracting CAC in the numerator and then interpreting the ratio as pre-acquisition value mixes conventions.
A common shortcut is periodic contribution divided by periodic customer churn. It comes from a highly simplified model, rather than from the definition of churn itself. To see its assumptions, consider a different hypothetical customer: the first month’s $80 net cash contribution arrives at month-end; each following month is renewed with constant probability r. There is no reactivation, expansion, contraction or change in contribution. Discount each month at d. Fader and Hardie show why this payment timing matters. [4]
The first contracted month’s contribution is assumed certain; later contributions depend on renewal. The expression sums m / (1 + d) + mr / (1 + d)2 + mr2 / (1 + d)3 + …. For nonnegative discount rates, it requires q + d to be positive. With d set to zero and q positive, it reduces to m / q. Receiving the initial contribution immediately or valuing only future renewals changes the formula. Zero observed churn does not establish an infinite customer value.
Assume m is $80 after delivery and retention costs but before acquisition cost, shared corporate overhead and taxes. CAC is $800 paid at acquisition. The following scenarios hold m and CAC fixed and change only the assumed churn or discount rate. They are sensitivity calculations, not estimates derived from the preceding cohort.
| Churn q | Discount d | Expected contribution LTV | LTV / $800 CAC |
|---|---|---|---|
| 2% | 0% | $80 / 0.02 = $4,000 | 5.00x |
| 2% | 1% | $80 / 0.03 = approximately $2,667 | 3.33x |
| 4% | 1% | $80 / 0.05 = $1,600 | 2.00x |
At 2% churn and a 1% discount rate, the value, rounded to cents, is $2,666.67, giving $2,666.67 / $800 = 3.33x. Doubling assumed churn reduces the value to $1,600 even with unchanged contribution. These are contribution values under the model, not company equity values or automatic spending limits.
Do not insert NRR into a customer-survival formula. Revenue retention includes account-size changes and can exceed 100%. Even aggregate customer retention does not establish a constant future survival probability for a particular acquisition cohort. Customer tenure, segment mix and changing behavior require separate analysis. [5]
Revenue per customer multiplied by gross margin is sometimes used as a contribution proxy, but it can omit customer-success, renewal and other retention costs. CAC also needs consistent treatment of sales compensation, marketing, onboarding allocations, unsuccessful acquisition efforts and the lag between spending and customer wins. Match customer segments and acquisition periods; dividing this quarter’s expense by unrelated customer additions can distort the ratio. There is no universal LTV/CAC threshold that proves a business is attractive.
How to Analyze Retention and Revenue Quality
- Read the definition before the headline. Record the cohort, unit, period, revenue basis and treatment of reactivation, currency and acquisitions. Preserve the issuer’s own label.
- Reconcile the opening cohort. Separate cancellations, contraction and expansion, then bridge separately to new and reactivated customers. Keep customer-count and dollar denominators distinct.
- Compare customer retention, GRR and NRR together. Investigate whether expansion is broad or concentrated in a few accounts and whether price changes, usage or additional products drive it.
- Track cohorts at comparable ages. Compare the first renewal of recent acquisition cohorts with the first renewal of older cohorts. A blended company average can change because the customer mix changes.
- Check the economics. Compare retained revenue with service costs, acquisition/expansion spending, margins and cash generation. Revenue retention does not measure profitability.
- Translate evidence into scenarios. Use retention and acquisition assumptions to inform revenue and cash-flow forecasts, with alternative outcomes. Keep the broader valuation mechanics in DCF and comparable-company analysis.
When assessing an EV/EBITDA multiple, retention can help evaluate the durability of the earnings behind the denominator. It does not dictate a warranted multiple. A business with excellent retention can still be overpriced, and contractual lock-in can postpone rather than eliminate renewal risk.
Prediction is a different task from measurement. Logit and probit models can estimate a defined future cancellation probability from customer characteristics; a reported historical churn rate simply summarizes outcomes under its measurement rules. Neither one, by itself, establishes why customers leave or whether a retention intervention will work.
Common Mistakes
- Offsetting lost customers with new customers. An increasing total customer count can coexist with weak opening-cohort retention.
- Calling every revenue loss churn without defining it. Distinguish full cancellation from contraction, and say whether reported revenue churn includes both.
- Using only NRR. Expansion can offset losses while GRR and customer retention weaken. Read all three measures and their underlying movements.
- Counting losses twice. Do not include a departed customer’s full opening revenue in both cancellation and contraction, or add expansion again to a closing-cohort numerator that already includes it.
- Mixing time periods or revenue concepts. A monthly churn input needs monthly contribution and discount rates; closing MRR is not the income statement’s period revenue.
- Turning an observed average into a survival forecast. One month of company churn, an annual NRR figure or a quiet retail customer does not establish an individual’s remaining lifetime.
Limitations of Churn and Retention Analysis
Measurement choices can change the result. Customer-account consolidation, acquired customers, grace periods and constant-currency adjustments can affect comparisons even without a change in underlying loyalty. Recalculate consistently where disclosure permits; otherwise state the comparison limit.
Renewal timing can delay the signal. A multi-year contract can keep a customer classified as active before the business faces the next renewal decision. Usage and payment evidence may add context, but neither should silently replace the published retention definition.
Retention does not identify cause or full value. A price increase can raise NRR today while influencing future cancellations. A renewal can require an expensive concession. These possibilities require evidence; a retention percentage cannot settle them. Customer cash-flow models must also allow for uncertainty, shared costs and reinvestment before being used in a business valuation.
Frequently Asked Questions
Sources
- Snowflake, Second Quarter Fiscal 2027 Results, September 2, 2026, financial highlights and Key Business Metrics.
- Workiva, Form 10-Q for the quarter ended September 30, 2026, MD&A operating metrics and retention definitions.
- U.S. Securities and Exchange Commission, Commission Guidance on MD&A: Key Performance Indicators and Metrics, Release 33-10751, effective February 25, 2020.
- Peter S. Fader and Bruce G. S. Hardie, Reconciling and Clarifying CLV Formulas, March 2012, cash-flow timing cases and model limitations.
- Peter S. Fader and Bruce G. S. Hardie, What’s Wrong With This CLV Formula?, December 2014, cohort dynamics and contractual versus noncontractual relationships.
- Stripe, Revenue churn: How to calculate it and why it matters, updated August 26, 2024, gross and net revenue churn.
Disclaimer
This article is for educational and informational purposes and does not constitute investment or accounting advice. Operating examples are hypothetical, and company metrics use the dated definitions cited. Lifetime-value calculations depend on assumptions and do not establish a business’s value or an appropriate acquisition budget.
