Churn Rate: How to Calculate and Reduce Customer Attrition

A software company closes 30 new contracts in a month and celebrates at the results meeting. That same month, 28 customers cancelled, and that number never made it onto a slide. By year's end, revenue is flat and nobody can explain why. That's the classic picture of a business that measures acquisition with a magnifying glass and customer attrition out of the corner of its eye.

Churn math is unforgiving because it compounds. A 5% monthly churn rate sounds manageable, but it means keeping only 54% of your customer base after twelve months. At 7% per month, more than half your customers are gone within a year. Few numbers erode a growth plan this fast, or this quietly.

Churn rate is the percentage of customers (or revenue) a company loses over a period. The formula: churn rate = (customers lost during the period ÷ customers at the start of the period) × 100. If you started the month with 400 customers and 28 cancelled, your monthly churn was 7%.

How to calculate churn rate in practice

The formula is simple, but three methodology decisions completely change the result. First: new customers acquired during the period don't belong in the denominator. If you started August with 400 customers, signed 50 and lost 28, churn is 28 ÷ 400 = 7%, not 28 ÷ 450. Mixing acquisition into the calculation hides the problem precisely when the company is growing fastest.

Second: define what counts as a "lost" customer. A formal cancellation is obvious, but what about the contract that expired and wasn't renewed? Or the customer inactive for 90 days in a non-subscription model? There's no single right answer; there's a consistent answer. Pick a criterion, document it, and stop moving the goalposts every quarter.

Third: keep the period fixed. Monthly and annual churn don't convert by multiplying by 12. At 5% monthly churn, the annual loss is 1 − (0.95)¹² = 46%, noticeably less than the naive "60%" a straight-line calculation would suggest, because the base shrinks every month.

Customer churn or revenue churn?

Counting heads is only half the story. Losing ten $40-a-month customers hurts less than losing one paying $3,000. That's why recurring-revenue businesses track both versions, and they tell different stories:

Metric	Formula	What it reveals
Customer churn (logo churn)	Customers lost ÷ customers at start × 100	Health of the base: how many accounts leave
Gross revenue churn	(Cancelled MRR + downgrades) ÷ starting MRR × 100	How much recurring revenue slipped away
Net revenue churn	(Lost MRR − expansion) ÷ starting MRR × 100	Whether upsells offset the losses

A round-number example. A company starts the month with $200,000 in MRR. It loses $9,000 to cancellations and $3,000 to downgrades, but existing customers expand by $15,000 in upgrades. Gross revenue churn is (9,000 + 3,000) ÷ 200,000 = 6%. Net churn is (12,000 − 15,000) ÷ 200,000 = −1.5%. Negative net churn: the existing base grows on its own, even while losing accounts. It's the scenario every subscription business chases.

In practice, revenue churn matters more for cash, and customer churn matters more for diagnosis. When gross churn rises but logo churn doesn't, the problem sits in your large accounts. When both rise together, the problem is structural: product, pricing, or the wrong audience. This kind of reading ties churn directly to LTV: retaining more is the cheapest way to raise the value of every customer.

What is a good churn rate? Benchmarks by industry

It depends on your business model, ticket size and contract cycle, and you should distrust anyone who answers without asking about those first. That said, some references help. The 2025 churn report from Recurly Research, which analyzes thousands of subscription businesses, puts the average monthly churn at 3.5%: 2.6% voluntary (the customer chose to leave) and 0.8% involuntary (failed payments).

For annual B2B contracts the yardstick changes. The State of B2B Account Experience study by CustomerGauge measured average annual churn of 14% in software, 19% in financial services and 31% in telecommunications. For SaaS on annual contracts, industry surveys put the median at around 10% per year or below.

An unpopular opinion: external benchmarks are for calibrating expectations, not for managing. The comparison that changes decisions is against your own history, broken down by signup cohort, acquisition channel and ticket band. A 4% monthly churn can be excellent for a $10 self-serve product and a disaster for a $1,000 contract with dedicated onboarding. Those cuts, by the way, belong among your core KPIs, not buried in a CS spreadsheet.

Which signals show up in the data before a cancellation?

Customers almost never cancel out of the blue; they cancel after weeks of signals nobody cross-referenced. The most common ones, in their typical order of appearance: declining usage (logins, sessions, key features), recurring support tickets with no definitive fix, the internal champion going quiet on the account, and finally financial signals such as late payment, a downgrade request or a declined card.

A plausible mini-case: a food distributor in Campinas, Brazil, selling on monthly contracts, notices that customers who reduce order volume for two consecutive months cancel within 90 days in 60% of cases. Just by creating an alert for "volume drop for 2 months" and routing it to sales, it starts fighting for the customer while there's still a game to play. No sophisticated model involved: a simple cross of orders per customer per month already delivers the warning. Teams that want to go further can use AI-based forecasting models to score each account's risk, but the simple rule captures most of the value.

How to reduce churn: what works first

Start with involuntary churn, the cheapest to attack. Within Recurly's 3.5% average, 0.8 percentage points come from failed billing: expired cards, insufficient funds, declined transactions. Smart payment retries, warnings before a card expires and alternative payment methods recover part of that without a single difficult conversation.

On voluntary churn, the highest-return lever is usually the start of the journey: customers who don't reach "first value" in the opening weeks rarely stay. Then come alerts on the signals from the previous section, annual contracts with an honest discount and, above all, real exit interviews. If 40% of cancellations cite the same issue, you don't have a retention problem; you have a product problem whose symptom is retention.

The prize for getting this right is large. A classic Bain & Company study by Frederick Reichheld showed that increasing retention by 5 points lifts profits by 25% to 95%, depending on the industry. And every retained customer is one you don't have to re-acquire by paying CAC all over again.

Frequently asked questions

What is a good churn rate?

As a general reference: subscription businesses average 3.5% per month (Recurly, 2025), and healthy B2B SaaS on annual contracts sits around 10% per year or below. The larger the ticket and the longer the contract, the lower your acceptable churn should be.

What is the difference between churn and retention?

They're complementary: retention = 100% − churn over the same period. A 7% monthly churn equals 93% monthly retention. Which one you display is communication; the information is the same.

What is negative churn?

It's when expansion revenue from customers who stay (upgrades, added seats, upsell) exceeds the revenue lost to cancellations and downgrades. MRR grows even without new customers. It only applies to revenue churn; customer churn is never negative.

How do you calculate revenue churn (MRR churn)?

Add the MRR lost to cancellations and downgrades in the period and divide by MRR at the start of the period. For the net version, first subtract expansion revenue from existing customers. Example: you lost $12,000 and expanded $15,000 on a $200,000 base: net churn of −1.5%.

What is involuntary churn?

It's losing customers to payment failure rather than an actual decision to leave: expired card, insufficient balance, transaction error. It typically accounts for about a fifth of total subscription churn and is the easiest slice to reduce, with retries and a dunning sequence.

Churn is a company metric, not a CS metric

As long as churn lives in a monthly report only the customer success team opens, it will remain an obituary: accurate, but late. The number that changes the game isn't last month's rate; it's the cross-reference between usage, support and billing that shows who is 60 days away from cancelling right now. That requires looking at CRM, product and finance together, which is more than almost any spreadsheet can take.

That's exactly the kind of question Sherlok answers: by connecting your CRM, spreadsheets and sales data, you ask in plain language "which customers reduced purchases for two consecutive months?" and get the list back, analysis included. Churn will stay compounding and unforgiving; the difference is seeing it before it becomes a statistic.