SaaS churn rate looks like one percentage. In practice, it can describe lost customers, lost recurring revenue or an avoidable billing failure. The useful work starts when the business separates those problems and connects each one to a decision.
01
Two SaaS companies can report the same churn and have different problems
Imagine two SaaS companies that both report 4% monthly churn.
The first loses eight small customers from a base of 200. Most leave during their first six weeks because they never complete onboarding.
The second loses one large account after a contract review. Its customer count barely moves, but the cancellation removes a much larger share of monthly recurring revenue.
The headline percentage is the same. The commercial problem is not.
SaaS churn rate measures the customers or recurring revenue lost during a defined period. Chargebee's glossary makes the first important distinction: a business can calculate subscriber churn or revenue churn. One counts customer relationships. The other measures the money attached to them.
That distinction matters because a blended churn figure can encourage the wrong response. The first company needs to examine customer fit, onboarding and time to value. The second needs to understand account health, renewal risk and whether one customer represented too much of the revenue base.
Before trying to reduce churn, define which churn you mean.
02
Customer churn counts lost accounts
Customer churn, sometimes called subscriber or logo churn, is usually calculated like this:
Customers lost during the period / customers at the start of the period x 100
If a SaaS product begins the month with 200 paying customers and eight cancel, customer churn is 4%.
New customers acquired during the month do not belong in the starting base. They can replace the eight logos on the total customer chart, but they do not change what happened to the group that existed on day one.
This is where simple spreadsheet calculations often become unreliable. A team subtracts the month end customer total from the month start total, then calls the difference churn. That movement may also include new sales, reactivations, free plan changes, pauses and customers who joined and left within the same month.
Agree what counts as an active paying customer, when a cancellation becomes effective and how pauses, trials, free plans and delinquent accounts are treated. Keep that definition stable so one month can be compared with the next.
Customer churn is useful when accounts are broadly similar in value. It becomes less informative when one customer pays £50 a month and another pays £5,000.
03
Revenue churn shows the size of the leak
Revenue churn measures recurring revenue lost through cancellation and, depending on the definition, contraction such as downgrades or fewer seats.
Suppose a company begins the month with £50,000 in monthly recurring revenue. Existing customers create three movements: £3,000 is lost through cancellations, £1,000 is lost through downgrades and £2,000 is gained through upgrades.
Gross revenue churn ignores the upgrades and measures the unfiltered loss:
(£3,000 cancellation + £1,000 contraction) / £50,000 x 100 = 8%
Net revenue churn includes expansion from the starting customer base:
(£3,000 cancellation + £1,000 contraction - £2,000 expansion) / £50,000 x 100 = 4%
The same movement can also be expressed as 96% net revenue retention. ChartMogul defines net revenue retention as starting recurring revenue plus expansion, minus contraction and churn, divided by starting recurring revenue.
These measures belong together. Net revenue churn shows the overall movement after existing customers expand. Gross revenue churn prevents healthy upgrades from hiding customers who left or reduced their spend.
| Metric | Simple calculation | What it helps answer | What it can hide |
|---|---|---|---|
| Customer churn | Lost customers divided by starting customers | Are customer relationships being retained? | Differences in account value |
| Gross revenue churn | Cancellation and contraction MRR divided by starting MRR | How much recurring revenue leaked before expansion? | Revenue gained from retained customers |
| Net revenue churn | Cancellation and contraction MRR minus expansion, divided by starting MRR | Did the starting revenue base shrink after expansion? | The underlying gross loss |
| Net revenue retention | Starting MRR minus losses plus expansion, divided by starting MRR | How much revenue remains from the starting cohort? | New business and differences between segments |
| Involuntary churn | Customers or revenue lost after payment failure | How much loss came from billing rather than a choice to leave? | Product dissatisfaction and poor customer fit |
New business should be kept outside these retention calculations. A strong sales month can make total MRR rise while the existing customer base is shrinking.
04
The measurement period changes the story
A monthly churn figure cannot be turned into an annual figure by casually multiplying it by twelve.
If the same 4% monthly customer churn continued, the remaining customer base would compound. The calculation is 0.96 raised to the power of 12, which leaves about 61.3% of the starting customers. That is approximately 38.7% annual churn, not 48%.
This is an illustration, not a forecast. Monthly churn changes, new cohorts behave differently and annual contracts create renewal patterns that do not fit neatly into a monthly cancellation rate.
Choose a period that matches the buying model. A self service monthly product may need weekly operational signals and a stable monthly churn measure. A B2B platform with annual contracts may learn more from renewal rates, gross revenue retention and cohorts aligned with contract anniversaries.
Do not compare a monthly plan with an annual plan as though they have had the same opportunity to leave. Segmenting by contract term and customer age produces a more honest view.
05
Voluntary and involuntary churn need different owners
A customer can leave because the product no longer earns its price. They can also disappear because a card expired or a bank declined a renewal payment.
The first is voluntary churn. It may point to poor fit, weak onboarding, missing value, service problems, pricing pressure or a competitor.
The second is involuntary churn. The customer did not make a clear decision to cancel, but the billing process failed to recover the payment.
Stripe's revenue recovery documentation separates failed payments, payments still in recovery and recovered payments. That is useful because the response is operational. Retry timing, reminders, payment method updates, clear account status and a sensible grace period may recover a customer who still wants the service.
Mixing the two kinds of churn produces a vague retention project. Product and customer success teams cannot fix an expired card with another feature. Billing automation cannot repair a product that customers have stopped using.
Put both figures on the same reporting route, then give each problem an owner.
06
Cohorts and segments expose the pattern
An average churn rate can improve while an important part of the business gets worse.
Imagine older annual customers remain loyal while a new monthly plan attracts many smaller accounts that cancel quickly. The blended customer count may look busy and the overall churn rate may move only slightly. A cohort view shows the newer customers falling away after the first renewal.
Useful ways to split churn include:
- sign up month or quarter
- plan and contract term
- customer size and recurring revenue
- industry or use case
- acquisition channel
- onboarding route
- product version or integration used
- time since activation
Keep each segment large enough to mean something. A 50% churn rate from a group of two customers is a conversation starter, not a trend.
Cohorts are especially useful because they compare customers at a similar point in their relationship. Month one churn may reveal poor qualification or onboarding. Churn after a pricing change may reveal a value problem. Losses at renewal may point to contract management, service reviews or competitive pressure.
07
Churn is a result, so look for earlier signals
Cancellation is usually the end of a story the product has been telling for weeks or months.
The useful question is not only which customers left. It is what changed before they left.
For a product used every day, falling completion of the core workflow may be an early warning. For quarterly compliance software, login frequency may be almost meaningless. Each product needs a definition of healthy use based on the result the customer bought.
Possible signals include:
- failure to complete onboarding
- long time to the first useful result
- reduced completion of the core workflow
- repeated integration or processing failures
- unresolved support cases
- fewer active teams, sites or seats
- a downgrade or paused account
- a failed renewal payment
- no engagement before a contract review
These signals should not become a mysterious health score that nobody can explain. Show the underlying behaviour and use it to decide who needs help, which part of the product needs attention or whether the customer was a poor fit from the beginning.
08
Cancellation reasons need evidence, not a dropdown alone
A cancellation form can collect useful feedback, but people choose the quickest available answer. One person selects "too expensive" because the product never became important enough to justify any price. Another selects "missing features" when the real problem was a difficult setup.
Use stated reasons beside observed behaviour, support history, account value and a small number of real conversations. Look for patterns rather than treating each dropdown selection as a diagnosis.
Make cancellation respectful. A confusing flow designed to trap people may delay a cancellation while damaging trust and creating support work. Offer a pause, downgrade or help route when it genuinely suits the situation, but keep the exit clear.
The goal is not to prevent every customer leaving. Some customers should leave because their needs changed or the product was never a sensible fit. Retention work should protect valuable, suitable relationships rather than hide poor acquisition decisions.
09
Reduce churn by fixing the cause in the right order
Once the churn is split into meaningful groups, the improvement work becomes much more specific.
If new customers leave before activation, reduce the number of steps between sign up and the first useful outcome. Clarify who owns setup, remove unnecessary data collection and make blocked accounts visible.
If customers stop using the core workflow, investigate whether the product still fits the job, whether a release made the route harder or whether the customer has moved the work elsewhere.
If larger accounts leave at renewal, review account concentration, service history, contract expectations, stakeholder changes and the evidence used to demonstrate value before renewal.
If failed payments create churn, improve retries, reminders, payment method updates and the route back into good standing.
If one acquisition channel produces unusually high early churn, revisit the promise, qualification and customer profile before spending more to fill the same leaking bucket.
Not every answer requires new software. A clearer onboarding call, a better renewal process or a change to qualification may be the smallest useful move. Build or integrate something when the repeated work, data and ownership genuinely need a system.
10
Churn reporting needs connected customer records
A dependable churn view often needs data from billing, CRM, product events, support and the application itself.
The difficult part is not drawing the percentage. It is agreeing which customer each record belongs to and what happened when. One company may have an account in CRM, several subscriptions in billing, multiple workspaces in the product and a different identifier in support.
Define the customer and subscription model before joining the numbers. Record the effective cancellation date, plan movement, reason, payment state and recurring revenue movement. Preserve enough history to understand the account before and after the event.
Then build a reporting route that moves from the headline to the evidence:
- Is churn materially different from the expected range?
- Is the movement in customers, gross revenue, net revenue or failed payments?
- Which cohort, plan or account type changed?
- What behaviour and events appeared before the loss?
- Who owns the next investigation or improvement?
That route is more useful than a dashboard full of disconnected retention charts.
11
Use churn as a decision system, not a verdict
Early SaaS products should be careful with percentages from tiny samples. One cancellation from ten customers creates 10% churn, but the right response is usually to understand the customer rather than redesign the whole product from one data point.
Larger products face the opposite problem. A stable average can hide meaningful losses inside a plan, country, use case or customer cohort. They need segmentation without turning every movement into noise.
The best churn reporting keeps both views available. It shows whether the recurring base is healthy, then makes the underlying customers and causes easy to inspect.
SaaS churn rate is not one answer. It is the start of a better set of questions about customer value, revenue quality, product use, service and billing.
If your churn reporting is spread across billing, CRM, product analytics and support tools, I can help define the measures, connect the customer records and shape a reporting workflow around the decisions your team needs to make.
Useful questions
Questions to settle before trusting a SaaS churn report
- Does churn mean lost customers, lost recurring revenue or both?
- Are new customers excluded from the starting retention cohort?
- Are downgrades, pauses, free plans and reactivations treated consistently?
- Can gross revenue loss be seen without expansion hiding it?
- Is involuntary churn separated from deliberate cancellations?
- Can the rate be split by plan, contract term, customer size and cohort?
- Are annual renewals compared on a suitable time period?
- Can cancellation reasons be checked against actual product and support behaviour?
- Does every important movement have an owner and a route into the underlying accounts?
- Are small sample sizes and changing definitions made visible?
