Unit Economics Wiki

Churn Rate

Churn rate measures the share of a frozen starting base — customers or recurring dollars — lost during a defined period, before any new sales are counted.

Unit EconomicsUpdated Aug 13, 202610 min read

Snapshot

What it is

The share of a frozen starting base lost over a defined window. Measured two ways that answer different questions — logo churn counts relationships lost, revenue churn weights each loss by its economic size.

Why it matters

Churn sets the floor your sales team must clear before growth begins, it is the single largest driver of lifetime value, and a small error in long-run churn produces a large error in LTV.

What it is not

It is not net growth, and it is not an audited number. New customers never reduce churn; they belong in growth. Churn and retention are unaudited operating metrics, not GAAP line items, so every company defines its own churn event, cohort, and denominator.

Key takeaways

  • Freeze the cohort first. Anyone who joined during the window cannot churn out of the starting base.

  • Report logo and revenue churn together. Losing 25 small accounts and losing 1 large one look identical in logo churn and nothing alike in revenue.

  • Never annualize by multiplying by twelve. Use 1 − (1 − m)^12. At 5% monthly, the correct annual churn is 46.0%, not 60%.

  • Segment before you act. A blended rate hides the concentrated risk that actually kills the plan.

What is churn rate?#

Churn rate is the proportion of a starting base — customers or recurring revenue — that is gone at the end of a defined period.

logo churn          = customers lost from starting cohort / customers at start
gross revenue churn = (churned recurring revenue + contraction) / starting recurring revenue

The two move independently. A company can lose many small accounts with modest revenue impact, or one anchor account with severe revenue impact. Reporting only one of them is a choice about which story to tell.

Subscription businesses generally track four related figures:

MetricNumeratorWhat it answers
Logo churn
Accounts lost from the starting cohort
How many relationships did we fail to keep?
Gross revenue churn
Full cancellations + downgrades
How many recurring dollars did we lose?
1 − gross revenue churn
What did we keep before any upsell?
Retained dollars plus expansion
Did the installed base grow on its own?

Before any of these mean anything, you have to define the churn event: contract termination date, effective service-end date, non-renewal after a grace period, or recurring revenue reaching zero. Pauses, delinquency, mergers, product migrations, and partial downgrades each need a written rule. A cancellation notice is not the same event as lost recurring revenue, and the gap between them is often a full quarter.

Why does churn matter to founders?#

Churn compounds, and it compounds first. Recurring revenue must replace losses before new acquisition produces net growth. At 3% monthly revenue churn, roughly 31% of the starting base has to be rebuilt each year just to stand still — before the growth plan starts.

It diagnoses fit by tenure, not in aggregate. Churn concentrated in the first 90 days points at targeting, onboarding, or an over-promised sale. Churn concentrated at renewal points at unrealized value, budget pressure, or a competitor. The operating response is completely different, so the blended number is close to useless as a diagnostic.

It changes acquisition economics more than anything else you control. The same CAC buys wildly different lifetime value under different survival curves, and because LTV models divide by churn, the error is non-linear: an error at low churn rates hurts far more than the same absolute error at high rates.

It determines what investors will underwrite. Diligence separates growth-from-new-logos from growth-from-the-existing-base. Reporting logo churn, GRR, NRR, and customer concentration side by side makes the revenue engine auditable — and the absence of one of them is read as a deliberate omission.

How do you calculate churn rate?#

1. Freeze the starting cohort#

List the accounts and their recurring revenue at the start of the window and do not touch it. Customers acquired mid-period are not eligible to churn from that cohort. Keep the membership list reproducible so the number can be recalculated months later.

2. Classify every movement#

MovementDefinitionCounts in gross churn?Counts in NRR?
Churn
Customer's recurring revenue falls to zero
Yes
Yes (negative)
Contraction
Customer stays, spends less
Yes
Yes (negative)
Expansion
Seats, usage, products, or price increase
No
Yes (positive)
New
Customer not in the starting cohort
No
No

3. Compute both views#

logo retention          = surviving logos from starting cohort / starting logos
logo churn              = 1 − logo retention
gross revenue retention = (starting RR − churn − contraction) / starting RR
net revenue retention   = (starting RR − churn − contraction + expansion) / starting RR

4. Align the window before comparing anything#

Monthly, quarterly, and annual churn are not interchangeable. Converting a constant monthly rate m to an annual figure requires compounding survival, not multiplication:

annual retention = (1 − m)^12
annual churn     = 1 − (1 − m)^12

5. Segment, then look at concentration#

Break the loss down by contract size, product, acquisition channel, tenure, use case, and geography — and always show the denominator and the recurring revenue at risk alongside the percentage. Then check whether one or two accounts drive the result.

6. Keep leading indicators separate from churn itself#

Product usage, support volume, champion departure, payment failure, and outcome attainment can predict churn. They are not churn. Validate any health score against realized renewals before you let it into a board deck.

Key Facts

01

Telecom "annual churn" is usually a monthly number

Rogers reported full-year 2025 postpaid mobile phone churn of 1.11% monthly (Q4 2025: 1.43%). Compounded, 1.11% monthly implies roughly 12.5% annual churn — not the 13.3% you get by multiplying by twelve.

Rogers Q4 2025 results
02

The denominator is a policy choice, and companies change it

Effective 1 April 2025 Rogers added 96,000 postpaid and 5,000 prepaid subscribers to its base from migrated brands, and excluded customers affected by its 3G decommissioning from its churn metrics entirely. Both moves are disclosed, and both change the reported rate.

Rogers Q4 2025 results
03

Best-in-class B2B SaaS revenue churn runs under 3% annually

Workiva reported a 97.2% gross retention rate and 112.8% net retention rate as of 31 December 2025 — roughly 2.8% annual gross revenue churn, with expansion adding 15.6 points on top.

Workiva FY2025 Form 10-K
04

Definitions are disclosed because they are not standard

Workday computes gross revenue retention by comparing prior-period ARR to current-period ARR for the same set of customers, counting product and customer churn but excluding add-ons and expansion, and reported approximately 97% as of 31 January 2026.

Workday FY2026 Form 10-K
05

A blended rate can hide the segment that matters

Similarweb reported overall NRR of 98% in Q4 2025, while NRR for customers with $100,000+ ARR was 103% — down from 112% a year earlier. Two different segments, two different problems.

Similarweb Q4 and FY2025 results

Worked example: one month, five numbers#

A company opens the month with 500 accounts and $250,000 MRR. During the month:

MovementAccountsMRR
Churned (starting cohort)
25
−$20,000
Contraction (survivors)
—
−$5,000
Expansion (survivors)
—
+$15,000
New customers
—
+$30,000

Logo churn

25 / 500 = 5.0%

Gross revenue churn (churn plus contraction, over the starting base)

($20,000 + $5,000) / $250,000 = 10.0%
GRR = 1 − 10.0% = 90.0%

Net revenue retention

($250,000 − $20,000 − $5,000 + $15,000) / $250,000 = $240,000 / $250,000 = 96.0%

Ending MRR

$250,000 − $20,000 − $5,000 + $15,000 + $30,000 = $270,000   (+8.0%)

Read those together: total MRR grew 8% in a month where the existing base shrank by 4%. New acquisition papered over the loss. Note also that revenue churn (10%) is double logo churn (5%) — the accounts that left were, on average, twice the size of the survivors. That is the diagnostic, and it disappears if you report only one number.

Why can't you multiply monthly churn by twelve?#

Because churn applies to a shrinking base. Each month's losses come out of what survived the previous month, so the annual figure compounds rather than adds. Using the 5% monthly logo churn above:

annual retention = 0.95^12 = 54.04%
annual churn     = 1 − 0.5404 = 45.96%   ≈ 46.0%

Not 60%. The naive figure overstates the loss by 14 percentage points, and the error grows with the rate:

Monthly churnNaive × 12Correct annual 1 − (1 − m)^12Overstatement
1%
12.0%
11.4%
0.6 pts
2%
24.0%
21.5%
2.5 pts
3%
36.0%
30.6%
5.4 pts
5%
60.0%
46.0%
14.0 pts
10%
120.0%
71.8%
nonsense → 48.2 pts

The reverse conversion has the same trap. A 30% annual churn rate implies 2.93% monthly (1 − (1 − 0.30)^(1/12)), not 2.50%.

Caveat on the formula. 1 − (1 − m)^12 assumes a constant hazard — that a customer in month 11 is exactly as likely to leave as one in month 1. Real churn is almost always front-loaded, so the constant-hazard projection understates early loss and overstates late loss. Use it to sanity-check a headline number, not to forecast. Wherever you can, observe an actual 12-month cohort instead.

What are the common mistakes?#

  • Letting new customers offset churn. Netting acquisition against loss produces a flattering number that measures nothing. Acquisition belongs in growth; churn belongs to the frozen cohort.
  • Mixing logos and dollars without saying which. State explicitly whether each customer or each recurring dollar carries equal weight. Investors will ask, and the gap between the two is itself informative.
  • Annualizing by multiplication. See above. This one shows up in seed decks constantly and is instantly recognizable.
  • Reporting a single blended rate. Segment, contract size, and tenure routinely hide a concentrated problem — one enterprise logo, one channel, one cohort.
  • Changing the churn definition without restating. Excluding involuntary churn, moving the effective date, or reclassifying a migration all shift the number. If the policy changes, restate the prior periods or quantify the effect.

When does churn rate break?#

Non-contractual and episodic purchases. For transactional businesses, inactivity is not termination — a customer who has not bought in 90 days may simply not need anything. Use repeat-purchase, survival, or probabilistic-inactivity models instead of forcing a churn definition onto the data.

Usage-based pricing. Revenue can fall substantially without any formal contraction event. Separate customer status from spending level, track contracted minimums apart from variable consumption, and use a smoothed baseline when seasonality is material.

Corporate actions. Mergers between two of your customers, customer acquisitions, plan migrations, and multi-workspace accounts all distort logo counts. Decide in advance whether consolidated entities, separate products, and separate workspaces count as one customer or several, and apply it consistently.

Churn tells you nothing about cause. A budget-driven non-renewal in a single vertical and a product-failure churn produce the same number and demand opposite responses. Always pair the rate with reason coding from actual renewal conversations.

Frequently asked questions

01

Should I lead with logo churn or revenue churn?

Revenue churn, with logo churn immediately alongside it. Revenue churn is what changes the financial model; logo churn tells you whether the problem is broad or concentrated. Reporting one without the other invites the assumption that you picked the flattering one.

02

Do I include involuntary churn from failed payments?

Include it in the headline number and break it out separately. Involuntary churn is real lost revenue, but it is fixed with dunning, card-updater services, and retry logic — not with product work. Companies that exclude it entirely are usually flattering the number.

03

What is a "good" churn rate?

It depends entirely on segment and price point. Public enterprise SaaS clusters around 2–5% annual gross revenue churn (Workiva reported 2.8%); self-serve SMB products routinely run 3–5% monthly. Compare yourself to companies with your contract size and buyer, never to a cross-industry average.

04

How does churn feed into LTV?

The simplest form is LTV = ARPA × gross margin / churn rate, using the same period for ARPA and churn. Note the shape of that formula: because churn is in the denominator, going from 3% to 2% monthly churn increases LTV by 50%. That sensitivity is why long-run churn assumptions deserve more scrutiny than almost any other input, and why you should always show an LTV range rather than a point estimate.

05

Are churn and retention numbers audited?

No. They sit outside the audited financial statements as operating metrics, and companies retain real discretion over the churn event, the cohort, the denominator, and what gets excluded. Rogers' 3G-decommissioning exclusion is a disclosed, defensible example — but it is still a choice. Read the definition before you benchmark against anyone.

Sources#

  1. Rogers Communications Inc., Fourth Quarter 2025 Results, published 2026. Primary disclosure of monthly postpaid and prepaid churn, the April 2025 subscriber-base adjustment, and the exclusion of 3G-decommissioning customers from churn metrics.
  2. Workiva Inc., FY2025 Form 10-K, for the year ended December 31, 2025. Gross retention rate of 97.2% and net retention rate of 112.8%, with the annualization methodology stated.
  3. Workday, Inc., FY2026 Form 10-K, for the year ended January 31, 2026. Gross revenue retention definition on an ARR basis for a fixed customer set, excluding add-ons and expansion.
  4. Similarweb Ltd., Fourth Quarter and Fiscal 2025 Results, February 17, 2026. Overall NRR of 98% versus 103% for customers with $100,000+ ARR, illustrating segmentation effects.

Source-use note: Company filings and results releases are primary statements by the issuer. They show how each company defines and reports its own retention metrics; they are not standardized and are not comparable without reading the definitions.

Note: This page is educational and does not constitute accounting, financial, or investment advice. Churn and retention are unaudited operating metrics that fall outside GAAP and IFRS, and companies exercise genuine discretion over cohort definition, denominators, exclusions, and restatement. Consult a qualified accountant or auditor before relying on these metrics in investor reporting or a financing document.

Author

Dr. Sarah Zou

Independent economist · EconNova

Commercial strategy for technical products, with a focus on pricing, unit economics, and the operating choices behind the model.

About Sarah

Topics

churn ratecustomer churnrevenue churnlogo churnretentionSaaS metricsunit economics

Cite this page

Suggested citation

Zou, S. (2026). Churn Rate: Measuring Customer and Revenue Loss Correctly. In Unit Economics. Pricing & Monetization Wiki. https://sarahzou.com/wiki/unit-economics/churn-rate

Open license

Reuse with attribution

This content is available for reuse. When referencing or republishing it, please credit Dr. Sarah Zou and link back to the original source.

Licensed under Creative Commons Attribution 4.0 International. You may share and adapt the material with appropriate credit.