The percentage of customers (or revenue) leaking out of your business each month — and the metric that quietly decides whether your SaaS compounds.
Research-based overview. Built from public benchmark reports (ChartMogul, Baremetrics, OpenView) and patterns we have seen working with solo founders at $0–$30k MRR. How we research.
Here is the formula in its plain form:
Five percent monthly may sound small. Compounded over a year — (1 - 0.05)12 = 0.54 — it means you lose 46% of your customer base every year just to churn. That is the math that makes churn the most consequential number in subscription businesses.
Three different measures of the same underlying phenomenon. They diverge because not all customers are worth the same.
The percentage of customers who left, regardless of how much they paid. Easy to calculate. Treats a $9 customer the same as a $900 customer.
Scenario: You start the month with 100 customers. Five cancel. Your customer churn is 5%. It does not matter if those five were on the cheapest plan or the most expensive — logo churn weights them equally.
The percentage of recurring revenue lost from cancellations and downgrades. This is the truer measure of damage to your business.
Scenario: Same 100 customers. The five who cancel were all on a $99 enterprise plan; the 95 who stay are all on a $9 starter plan. Customer churn is 5%, but revenue churn is (5 × $99) / (5 × $99 + 95 × $9) = $495 / $1,350 = 36.7%. You are losing more than a third of your revenue while your “customer churn looks fine” dashboard says 5%.
NRR adds expansion back in. Take the MRR from a cohort of customers at the start of a period, then look at what those same customers contribute at the end of the period — including upgrades, seat expansions, and add-ons but minus their churn and contraction.
Scenario: The cohort started at $10,000 MRR. By month-end, $1,200 had churned, $400 had downgraded, but $2,000 had upgraded or added seats. NRR = ($10,000 - $1,200 - $400 + $2,000) / $10,000 = 104%. NRR above 100% is “negative net churn” — the cohort grows even if no new customers join. Public SaaS companies with NRR above 120% (Snowflake, Datadog historically) are the ones with the wildest valuations.
Churn benchmarks depend almost entirely on what kind of SaaS you sell and to whom. Comparing yours to a number from a different segment is how founders end up panicking unnecessarily — or feeling falsely safe.
| SaaS type | Typical monthly churn | Why |
|---|---|---|
| B2C consumer SaaS | 5–7%/month is normal | Low switching cost, individual decision, used until it isn’t needed |
| B2B SMB | 3–5%/month | Some workflow embedding, but small businesses themselves churn at a baseline |
| B2B mid-market | Under 1%/month | Annual contracts, embedded in workflows, hard to rip out |
| B2B enterprise | 0.3–0.7%/month | Multi-year contracts, deep integration, procurement friction |
These are rough averages from ChartMogul’s SaaS benchmarks, Baremetrics’ published reports, and OpenView’s annual SaaS benchmark study. They will move year to year, but the relative ordering is stable: the more enterprise the buyer, the lower the churn ceiling that’s considered “normal.”
The dirty secret of churn metrics at small scale: they are mostly noise. There are three specific ways small-sample churn misleads founders.
If you have 30 paying customers and one cancels, your monthly churn rate is 3.3%. If two cancel, it is 6.7%. The metric just doubled and it tells you nothing about whether your business changed — you are watching a coin flip. Until you have at least 100 customers, individual churn events should be analyzed qualitatively (talk to the customer) rather than statistically (chart the rate).
SaaS churn is rarely flat across the year. B2C tools churn higher in January (post-holiday subscription cleanup) and August (vacation cancellations). B2B tools churn higher around fiscal year-end when budgets are reviewed. Looking at one month in isolation and concluding “churn is up 30%” without controlling for seasonality is a common mistake.
When a customer cancels on day 5 of their billing cycle, are they churned in this month or next? Stripe’s default is “they remain active until period_end,” which means they show as a customer for another 25 days even though they have decided to leave. Track cancellation intent separately from cancellation effect, or your churn metric will lag reality by an entire month.
For the broader set of operating metrics that matter at small scale — and the trade-offs of each — see our SaaS metrics that matter piece. PostHog can help instrument the cancel-flow events that feed churn analysis cleanly; we covered the trade-offs in our PostHog review.
Churn-reduction advice is mostly platitudes. Here are the five interventions that genuinely move the metric for a solo founder, ordered by leverage.
Mostly no. Below ~50 customers, churn is too noisy to draw conclusions from. Spend your time on activation and customer interviews; obsess over churn once you have enough data points for the rate to mean something.
Close, but not quite. Annual churn = 1 - (1 - monthly_churn)12. At 5% monthly that is 46%, not 60%. The compounding matters because each month’s churn applies to a smaller base.
Yes, eventually — this is “involuntary churn.” Most analytics tools default to counting a customer as churned once their card has failed and dunning has been exhausted (typically 14–21 days after first failure). Track involuntary and voluntary churn separately because the fixes are different (dunning emails vs product issues).
Negative net revenue churn means a cohort’s expansion (upgrades, seat additions) exceeds its cancellations and contractions. The logo count may still shrink, but the dollars per remaining customer grow faster than the loss. It is the holy grail of subscription metrics because the business compounds even with zero new acquisitions.
Churn rate is one of the few metrics in SaaS that can quietly kill a business while every other dashboard looks healthy. The fix is not chasing a single headline number but understanding which kind of churn you are measuring (customer, revenue, NRR), benchmarking against businesses like yours rather than the public-company averages, and resisting the urge to read signal out of noise at small sample sizes. Below 100 customers, talk to people. Above it, instrument and improve.
The stack, prompts, pricing, and mistakes to avoid — for solo founders building with AI.