Churn and attrition describe the same outcome, customers leaving, but teams use the two words for different things. Churn is the discrete event: a subscription cancels on a date you can point to, and you count it. Attrition is the gradual erosion around that event, including customers who never formally cancel and simply stop buying. Subscription businesses mostly say churn. Businesses with no cancel button, like retail, banking, and professional services, mostly say attrition.
Last updated: July 2026.
The reason this matters is not vocabulary. It is that the two words point at different measurement problems. If your customers cancel a contract, you have a clean event to count and the hard part is explaining it. If your customers just quietly buy less every quarter and one day stop, there is no event at all, and the hard part is noticing. Teams that use the words interchangeably usually end up measuring only the first kind of loss and missing the second entirely.
What is the difference between churn and attrition?
Churn is the recorded exit: a cancellation, a non-renewal, a closed account on a specific date. Attrition is the decline in customer value that leads there, including customers who fade out without ever cancelling. Churn is a binary event you count. Attrition is a slope you watch. In contractual businesses the two collapse into the same number. In non-contractual ones they do not.
Most reference sources and most SaaS blogs treat the terms as synonyms, and in casual use they are. TechTarget's definition literally reads "customer churn (customer attrition)." But practitioners who work in both worlds keep a distinction, and it is a useful one:
- Churn is an event. There is a timestamp. A subscription ends, a contract lapses, an account closes. You know exactly which customers churned in June because the billing system says so.
- Attrition is a trend. Order frequency drops from monthly to quarterly. Seats fall from 40 to 18. The client that used to send three projects a year sends one. Nobody cancelled anything. Value is leaving anyway.
A bank does not get a cancellation notice when a customer moves their direct deposit somewhere else and leaves $12 in the account. A law firm does not get one when a client stops calling. Those are attrition. Counting only formal account closures in those businesses understates the loss badly, sometimes by a factor of two or three, because dormant accounts sit on the books looking like customers.
Churn vs attrition vs turnover vs retention: side by side
| Term | What it counts | Typical business | Measured as | Signal timing |
|---|---|---|---|---|
| Churn | Customers or revenue lost in a period through a recorded exit | Subscription, SaaS, telecom, insurance | Lost customers or MRR divided by the starting base | Lagging. The decision was made weeks earlier. |
| Attrition | Erosion of customer count, spend, or activity, cancelled or not | Retail, banking, professional services, healthcare | Same math, applied to a definition of "active" rather than "subscribed" | Leading, if you define inactivity well. |
| Turnover | Almost always employees leaving, not customers | Any (this is an HR metric) | Separations divided by average headcount | Not a customer metric at all. Do not mix it in. |
| Retention | The customers or revenue you kept | All | The complement of churn, when both use the same base and window | Same timing as churn, framed positively. |
Are churn and attrition the same thing?
As an outcome, yes. Both mean you lost a customer. As a measurement practice, no. Churn assumes an exit event exists to be counted, which is true in subscription businesses and false in most others. Attrition covers the loss even when no exit event is ever recorded, which is why industries without contracts almost always use that word instead.
The tell is whether you can answer this question: "how many customers left last month?" If your billing system answers it exactly, you are measuring churn. If you have to first decide what counts as gone, you are measuring attrition, and that definition is now the most important decision in your reporting.
Employee churn vs attrition vs turnover
This is where the words genuinely collide, and it is worth being explicit because search results mix the two constantly. In HR, turnover counts everyone who left and got replaced, while attrition traditionally means departures where the role was not backfilled and headcount shrank. "Employee churn" is a borrowed term that usually just means turnover.
None of that transfers to customers. If you are reading a benchmark that quotes a 15% attrition rate, check which population it counts before you compare yourself against it. A 15% employee attrition rate and a 15% customer attrition rate are unrelated numbers that happen to share a word.
Churn rate vs attrition rate: how each is calculated
The arithmetic is the same. What changes is the numerator's definition.
Churn rate = customers lost during the period divided by customers at the start of the period. If you began January with 400 accounts and 18 cancelled, that is 18 / 400 = 4.5% monthly logo churn. Our walkthrough of the churn rate formula covers the edge cases, including what to do with customers who joined and left inside the same month.
Attrition rate = customers who became inactive during the period divided by the active base at the start. The formula is identical. The work is in the word "inactive." A gym might define it as no visit in 60 days. A distributor might define it as no order in two full purchase cycles. An accounting firm might use no engagement in 14 months, because annual work has a long natural gap.
Pick the inactivity window using your own repurchase data, not a round number. If 90% of repeat orders arrive within 75 days, then 120 days of silence is a real signal and 30 days is noise.
Worked example on the attrition side. A B2B supplier starts the quarter with 1,200 accounts that ordered in the prior six months. By quarter end, 96 of them have gone two full cycles without an order. Attrition rate is 96 / 1,200 = 8% for the quarter, roughly 28% annualized if the pattern holds. Not one of those 96 accounts cancelled anything, and none of them would appear in a churn report built on cancellations.
Gross churn vs net churn
Once you move from counting customers to counting dollars, a second split appears, and it is the one that most changes how a business looks.
- Gross revenue churn counts only revenue lost: cancellations plus downgrades. It cannot go below zero. It answers "how leaky is the bucket?"
- Net revenue churn subtracts expansion revenue from existing customers before dividing. It can go negative, which means your remaining customers grew more than your lost ones cost you.
Example. You start the month with $500,000 MRR. You lose $30,000 to cancellations and downgrades and gain $22,000 from upgrades and seat additions inside the existing base. Gross revenue churn is 6%. Net revenue churn is ($30,000 minus $22,000) / $500,000 = 1.6%. Same month, same customers, two numbers that tell very different stories. The mirror image of net churn is net revenue retention, which is the metric most investors actually ask for.
The trap: reporting net churn alone hides a retention problem while expansion is strong. Plenty of companies with healthy net numbers were bleeding underneath and only found out when expansion slowed. Report both, always.
Churn rate vs retention rate
Retention rate is what churn leaves behind, but only when both are measured on the same base over the same window. Start with 400 accounts, lose 18, and logo retention is 95.5% while logo churn is 4.5%. Those two always add to 100%.
Revenue is where the symmetry breaks. Net revenue retention can exceed 100% because expansion is counted, so it is not simply "100 minus churn." Gross revenue retention is the one that pairs cleanly with gross revenue churn. Mixing a net retention figure with a gross churn figure on the same slide is one of the most common reporting errors in SaaS, and it usually flatters the deck.
Churn and attrition benchmarks for 2026
Benchmarks are only useful against a like-for-like definition, so treat these as ranges to orient by rather than targets. SaaS Capital's 2026 survey of more than 1,000 private SaaS companies puts median net revenue retention at 103% and median gross revenue retention at 91% for bootstrapped companies between $3M and $20M ARR. The 90th percentile reaches 117.9% net and 100% gross.
| Segment | Typical annual gross churn | What drives it |
|---|---|---|
| SMB and self-serve, under $10K ACV | Roughly 18% to 32% | Small customers go out of business, switch cheaply, and buy without a committee |
| Mid-market | Roughly 8% to 16% | Annual contracts, some procurement friction, real but survivable switching cost |
| Enterprise, $200K and up ACV | Roughly 4% to 10% | Multi-year terms, integrations, and internal cost of change |
Two cautions. First, published 2026 compilations vary widely because they mix logo churn with revenue churn and monthly with annual figures. Confirm which one a number is before you quote it internally. Second, a monthly rate compounds: 3% monthly is not 36% annually, it is about 30.6%, because each month churns a smaller remaining base.
How to spot attrition before it turns into churn
The whole practical value of separating the two words is that attrition is visible earlier. The signals that reliably precede a cancellation, in rough order of how early they appear:
- Usage narrows to one person. The account still logs in, but only the original champion does. Breadth of use falls before depth does.
- The champion leaves. A departing internal advocate is one of the strongest single predictors of a lost account, and nothing in your product tells you it happened.
- Support contact changes shape. Fewer questions about how to do more, more questions about billing, exports, and data ownership. Export requests in particular sit very close to a cancellation.
- Invoices start getting queried or paid late. Payment behavior degrades before the relationship formally ends, which is why AR aging is quietly a retention report.
- Meetings stop getting accepted. The quarterly review slips twice and then disappears.
Most of these are operational rather than sentimental, which is the point made at length in measuring customer experience operations: the numbers that predict churn come from your systems, not your surveys. Rolling them into one weighted customer health score is how most teams turn the list into something an account owner can act on before renewal week.
One structural fix deserves its own mention, because it addresses the first two signals directly. Accounts churn hardest when only one person at the customer ever learned the product, so a single resignation takes the whole relationship with it. Teams that solve this stop treating training as a kickoff event and run it as a standing program, usually through a shared training and certification platform so new hires on the customer's side get up to speed without another live session being booked. Depth of trained users is a retention asset that shows up nowhere on a satisfaction survey.
What does a higher churn rate indicate?
A rising churn rate usually indicates a value problem that started at the beginning of the relationship, not the end. In most B2B businesses the strongest correlate of churn is failure to reach first value during onboarding. Pricing and competition matter, but they typically explain a smaller share of cancellations than never getting properly started does.
Before concluding that customers dislike the product, split the number. Separate voluntary churn from involuntary churn caused by failed card payments, which is a plumbing problem that a working dunning process recovers most of. Then split by cohort. If churn is concentrated in accounts onboarded in a particular quarter or by a particular team, you have a process defect with a date on it, not a market verdict. Tracking time to value and activation is how you find that.
What is a good churn rate?
For B2B SaaS, roughly 5% to 7% annual logo churn is strong, and under 1% monthly is generally considered good for companies selling to businesses. Self-serve and SMB products routinely run higher and are not automatically unhealthy, because their acquisition cost is lower. The honest answer is that churn is only good or bad relative to what a customer is worth.
That is why the number to judge yourself on is not churn alone but churn against customer lifetime value and acquisition cost. A 3% monthly churn rate on a product with a 6:1 LTV to CAC ratio is a working business. A 1% monthly rate on a 1.5:1 ratio is not.
What is the customer attrition rate formula?
Customer attrition rate = (customers who became inactive during the period / active customers at the start of the period) x 100. It is the churn formula with "inactive" swapped in for "cancelled." The formula is trivial and the definition of inactive is everything, so write that definition down, date it, and do not quietly change it between quarters. A redefinition can move the reported rate more than any real change in the business.
Which word should your team use?
Use churn if you sell subscriptions and your billing system records exits. Use attrition if customers can drift away without telling you. Use both, defined separately, if you have contracted and transactional revenue in the same business, which is common in services firms running retainers alongside project work.
What matters more than the label is that whichever number you report is defined once, in writing, with the window and the base stated. Nearly every argument about whether churn is improving turns out to be an argument about definitions. Once the number is stable, the work moves to the operational side of the house, which is where the retention strategies that actually hold in B2B live.