Churn rate formula and interpretation
Contents:
What churn is and why it matters
Churn is the mirror image of retention. Retention answers "how many stayed?" — churn answers "how many left?" Same population, opposite framing, and most subscription businesses live or die on which side of that ratio they obsess over. If you only watch one of the two, you will eventually miss something the other was screaming about.
Why it matters: subscription businesses break on churn long before they break on acquisition. You can double your paid acquisition budget at Netflix, Spotify, or any consumer SaaS and still shrink, because outflow grew in lockstep. Churn is a tax on growth — while you pour new users into the top of the bucket, someone is leaking through a hole at the bottom. If the hole widens faster than the pour, the company is shrinking with a straight face.
What leadership wants is not the single number — it is the trend. A steady 5% monthly churn is calmer news than a jagged 3% → 8% → 4% sequence. Jumps mean something broke: a price change, an app release, a billing policy update, or a new acquisition channel pulling in lower-quality cohorts. The level tells you the business model; the variance tells you what shipped recently.
The base churn rate formula
Churn rate = users who left during period / users active at start of period × 100%Period can be anything — day, week, month, quarter, year. Most subscription products report monthly churn, while annual-contract B2B SaaS reports annual churn (also called logo churn at that timescale).
Worked example. On April 1, a streaming product has 10,000 paying subscribers. Across April, 450 cancel. Monthly churn = 450 / 10,000 = 4.5%. Simple as that — until you start arguing about what goes in the denominator.
Be careful with the denominator. Some teams use the average of start and end of period — (start + end) / 2 — others use just start. The difference is invisible at scale but material on small books. In interviews, ask which convention the company uses before you defend a number. The same raw event log can produce two different churn rates depending on this choice alone.
The other ambiguity is what "left" means. In a paid subscription, the rule is mechanical: the customer canceled, or the renewal failed. In freemium products it gets philosophical fast — is "didn't open the app for 30 days" churn? If they open it on day 35, did they un-churn? This is why freemium teams usually report N-day inactive rate instead of churn, and reserve the word "churn" for paid product lines.
Customer churn vs revenue churn
Customer churn counts heads. Revenue churn counts dollars those heads were paying you. The two diverge whenever pricing tiers exist, which is roughly every interesting SaaS company.
Customer churn = lost customers / customers at start of period
Revenue churn = MRR of lost customers / MRR at start of periodIf a single enterprise account paying $8,000/month leaves, customer churn ticks up by one — revenue churn opens a crater. Conversely, if 100 self-serve accounts at $15/month each cancel, customer churn looks alarming while revenue churn barely moves. Neither view is wrong; they answer different questions.
For B2B SaaS with tiered pricing, revenue churn is the metric that matters more. The business runs on money, not on logo count. For consumer apps with a flat price, customer churn is usually fine because every cancellation is worth the same. The general rule: if your average revenue per customer varies by more than 5x across the book, you need both views side by side or you will mislead yourself.
| Metric | Counts | When it lies | When to trust it |
|---|---|---|---|
| Customer churn | Logos / accounts | Tiered pricing with large variance | Flat-price consumer apps |
| Revenue churn (gross) | MRR / ARR lost | Hides logo concentration risk | Tiered B2B with mixed segments |
| Net revenue churn | MRR lost − expansion | Masks weak new-logo motion | Mature SaaS with healthy expansion |
Gross churn vs net churn
Gross churn counts only the outflow. How much MRR or how many customers walked out the door — full stop.
Net churn subtracts upgrades, seat expansion, and cross-sell from the customers who stayed. It is the metric a CFO wants on a board slide.
Net revenue churn = (MRR lost − expansion MRR from existing customers) / MRR at start of periodWhen expansion from existing customers exceeds gross churn, net revenue churn goes negative. This is the famous "negative net churn" or NDR > 100% — you are losing customers, yet MRR from the remaining book is growing on its own. It is the holy grail of mature SaaS economics: even with the new-business pipeline frozen, the company still grows.
Load-bearing trick: If you take one thing from this post, it's that gross and net churn answer different questions. Gross says "are we keeping customers?" Net says "are we growing inside the accounts we already have?" A healthy SaaS report shows both — net alone can hide a leaky logo motion, gross alone can hide a brilliant expansion motion.
In product-role interviews, expect questions about negative net churn. Whoever asks is checking whether you understand subscription economics beyond top-line metrics, not just whether you can recite a formula.
How to interpret the number
The raw number says almost nothing. You need three pieces of context before any churn figure is meaningful.
Period. A monthly churn of 5% and an annual churn of 5% live in different universes. Naively, 5% per month compounds to ~46% over a year — and actually higher, because the base shrinks each month, which exposes a relatively larger slice to risk in absolute terms. Always pin the period to the number. A churn figure with no period attached is noise.
Segment. A global product churn rate can hide that one segment churns at 1% while another churns at 30%. Cut by plan tier, acquisition source, billing cycle, geography, and tenure. The averaged number is almost always less actionable than the segment view.
Customer lifecycle stage. Fresh customers leave faster than tenured ones — a well-known shape called the J-curve of churn. If your acquisition is growing, your blended churn will drift up not because the product got worse, but because the cohort mix shifted toward newer users who churn at higher base rates. Always look at churn within a cohort, not just blended.
Causation. Churn moved — why? Pricing, UX release, billing retry policy, traffic source shift, a competitor running a promo. Without a story, the metric is a thermometer with no diagnosis attached. Movement without explanation gets you a follow-up meeting, not a decision.
Benchmarks by segment
Any number below is a public-source ballpark, not a target. Categories differ wildly.
| Segment | Typical monthly churn | Typical annual churn | "Excellent" |
|---|---|---|---|
| B2B SaaS, annual contracts | n/a (logo) | 5–10% | < 5% |
| B2B SaaS, monthly plans | 3–7% | 30–55% | < 3% |
| Consumer subscriptions (streaming, fitness) | 5–8% | ~50% | < 4% |
| Mobile subscription apps | 8–15% | 60–80% | < 8% |
| Freemium → paid conversion churn | 15–25% (first 90d) | n/a | < 12% |
If your monthly churn is north of 20%, the average customer lives shorter than 5 months. Fixing retention beats spending more on acquisition — a leaky bucket leaks at every size.
Sanity check: Before celebrating a churn drop, ask whether the denominator changed. A flood of new free-trial users you reclassify as "active" lowers the ratio without anyone actually retaining better. Half the apparent churn improvements I have seen in interview portfolios were denominator artifacts, not retention wins.
Common pitfalls
The first trap is computing daily churn for a long-lived product. If average customer lifetime is a year, daily churn is around 0.3% and impossible to interpret meaningfully — noise dominates signal. Match the reporting cadence to the product's natural cycle: monthly for monthly-billed products, quarterly for annual contracts. Picking a tighter cadence than the product justifies makes every weekly review a coin flip.
Second is ignoring voluntary vs involuntary churn. A meaningful share of cancellations is not "I dislike this product" — it is "my credit card expired and the retry logic gave up after one attempt". This is fixed with dunning email reminders, smart retry windows, and account updaters, not with product changes. Mixing the two hides the fact that 20–40% of gross churn in many consumer subscription products is mechanical and recoverable.
Third is comparing churn across business models. A B2B SaaS with annual contracts and a mobile app with monthly subscriptions are different planets. Putting both in the same benchmark slide is the analyst equivalent of comparing the GDP per capita of a city-state and a continent. Always restrict comparisons to the same business model, billing cycle, and customer segment.
Fourth is never separating customer churn from revenue churn. You can celebrate that "only 2% of customers left" and miss that those 2% included your three largest contracts. The reverse also happens — a panic over "20% of accounts gone" when those accounts represented under 3% of MRR. The two numbers must be read together.
Fifth is counting trial-month churn as product churn. If a cohort just finished a 14-day free trial, most of them will not convert — that is by design, and folding them into the active base inflates the next month's churn dramatically. Separate trial conversion rate from post-trial churn, and you will stop having mysterious monthly spikes that line up with marketing campaigns.
Sixth is treating negative net churn as a nice-to-have. For mature SaaS, expansion revenue from the existing book is the cheapest growth source you have. Ignoring it means you grow only as fast as new-logo acquisition lets you — which, given typical CAC, is much slower than your competitors who land-and-expand. Net dollar retention above 110% is what separates a great SaaS from a fine one.
Related reading
- Churn explained simply
- How to calculate churn in SQL
- How to calculate cohort retention in SQL
- How to calculate involuntary churn in SQL
- How to calculate gross dollar retention in SQL
- Churn prediction modeling guide
If you want to drill metric questions like this every day with feedback, naildd ships 500+ product analytics and SQL problems across exactly this pattern.
FAQ
How is churn related to retention?
In the simplest one-period framing, they sum to 100% — retention 95% means churn 5%. In practice, formulas drift across companies because of denominator choices and how they handle returning users, partial-month customers, and trial cohorts. Always check the definition document before benchmarking your number against someone else's.
What is negative net churn and why do investors care?
Negative net churn is when expansion revenue from existing customers exceeds the revenue lost to cancellations and downgrades in the same period. It means the existing book is growing on its own, independent of new logos. Investors love it because it implies a compounding revenue base with structural pricing power — even a hiring freeze on the new-business side leaves growth intact for several quarters.
How do I actually reduce churn?
Fix onboarding so new users hit the activation moment in the first session, instrument early disengagement signals (sessions dropping, key feature usage falling) and act on them with lifecycle messaging, and tighten billing operations — retry logic, card updater services, and dunning windows. The product changes matter, but in many subscription businesses the billing fix alone recovers a meaningful share of gross churn before you touch the product.
What is voluntary vs involuntary churn?
Voluntary churn is when a customer actively cancels — they saw the bill, decided it was not worth it, hit the button. Involuntary churn is when a payment fails for mechanical reasons: an expired card, a hit fraud rule, a network timeout. Involuntary churn is cheaper to fix and faster to win back, which is why mature subscription businesses run a dedicated payments-recovery team well before they touch the product roadmap.
What is a "healthy" churn rate for a mobile subscription app?
Public benchmarks place healthy monthly churn for consumer mobile subscriptions in the 5–15% band, with category leaders below 8% sustained over multiple quarters. The number depends heavily on the use case (daily habit apps churn less than seasonal ones), price point, and onboarding quality. Comparing against your own segment is more useful than chasing an absolute number.
Can you compute churn for a free product?
You can, but you have to define "left" yourself — usually as "did not perform key action in the last N days". This is closer to an inactive rate than to subscription churn, and it behaves differently statistically. Two users who go dark for 30 days each may both return next month; in a paid model they would each count as one cancellation. Pick the threshold deliberately and keep it stable, otherwise quarter-over-quarter comparisons are meaningless.