You typed “good churn rate for early SaaS” because someone canceled and the percentage looked scary. With fifteen paying customers, one leave is not a trend. It is one story. The internet will still throw enterprise targets at you as if you ran a $20M ARR company.
Public medians exist. They come from companies with enough volume that the math settles. Indiecity is for one person or a team under twenty. Your job is to read those medians as context, refuse invented “you must hit X%” rules, and spend the week on who left and why.
What churn actually measures
ChartMogul’s customer churn guide keeps the split simple. Customer churn (logo churn) is how fast accounts leave. Revenue churn is how fast dollars leave. The same month can look fine on account count and ugly on dollars if a large account cancels.
The basic customer churn idea: customers who left in the period, divided by customers at the start of the period. ChartMogul’s published formula also strips same-period join-and-leave noise so a day-one trial cancel does not pretend to be a long-term account. Use one definition and stick to it. Do not change the formula every time the number looks bad.
ChartMogul’s own walkthrough is three customers at $20, $30, and $50. One cancel of the $50 account is 33% customer churn and 50% revenue churn in the same month. That is why both numbers matter, and why a tiny base makes both of them jump. If cash and the dashboard disagree, here is which MRR number to watch.
A small list makes the percentage jump
With 10 customers, one cancel is 10% monthly customer churn. With 25, one cancel is 4%. That jump is sample size, not product-market fit flipping overnight. Until the base is large enough that one leave does not rewrite the percentage, the rate is a weak grade. It is still useful as a prompt: who left, after how long, and for what reason.
Do not invent a personal target (“we need under 3% by month four”) to soothe the anxiety. If a number is not on a public primary page or on your own ledger, it is not a law for your product.
Read the public medians. Do not copy them as a weekly grade
On customer churn, ChartMogul’s median early-stage company under $300k ARR shows about 6.5% monthly customer churn. Cheap plans lose more accounts: their median sits near 6.1% when ARPA is under $25, and near 2.2% in the $500 to $1k band. Their writeup also shows that 5% monthly compounds to roughly 46% over a year if nothing changes. Useful for intuition at scale. Not a grade when you have twelve paying accounts.
In 2022, Lenny Rachitsky published monthly churn bands built with Patrick Campbell and ProfitWell’s set. Those “good / great” bands describe mature products in each segment, not a weekly score for a product still learning who buys. First-month churn is steeper. Your month-two product will not look like a five-year book. Stop comparing them.
Why people leave
ChartMogul names three product reasons customers cancel: they never reached value, price and value never met, or they were the wrong customer. That is the work, not a hunt for a prettier percentage. Onboarding with no success team is the first leak. Selling the wrong buyer is the second.
They also split voluntary cancels from delinquent ones: expired cards, failed charges, a payment that never updated. ChartMogul has seen cases where failed payments are as much as a third of churn. That pile is a billing job. It is not a verdict on product-market fit. How to reduce churn with twenty customers starts there: failed cards first, then the exit call.
How to calculate without lying to yourself
- Pick a period (usually a calendar month) and freeze the definition.
- Count paying customers at the start. Count who canceled or failed permanently in that period.
- Customer churn ≈ leavers ÷ starters. Keep same-month join-and-leave out if you follow ChartMogul’s published approach.
- Separately track revenue lost to cancel and downgrade (gross), and net of expansion if you have upgrades.
- Tag each leave: never activated, wrong fit, price, competitor, card failed, company dead, job done.
If you only have a spreadsheet, that is enough. A dashboard does not create retention. It only reports it. ChartMogul’s published example is 100 customers at the start of the month, 12 cancels, one reactivation, one same-period join-and-leave: (12 − 2) ÷ 100 = 10%. Write your own line the same way. Do not skip the subtractions to make the month look calmer.
What to do this week (not invent a target)
- Write the last 90 days of cancels with one reason each. Patterns beat percentages.
- Fix failed payments before you rewrite the product. Delinquent churn is real and often large.
- Watch day-1 to day-14 activation. If people never reach the outcome, they will not renew.
- Stop selling hard to the wrong segment. Wrong-fit revenue churns first.
- Reply fast when a real customer is stuck. You do not need a success team to answer email.
- Revisit price only if leavers say value and price never met. Do not cut price to bribe a bad fit to stay.
Write this for every cancel in the last 90 days. One card per person. Do not average them first.
Name / company: Paid for how long: Last time they used the product: They left because: [never finished setup / wrong buyer / price / competitor / card failed / company closed / job done] What I will do this week: [email the failed card / fix the setup step / stop selling to this segment / nothing, this leave was clean]
If you still have fewer than ten people who pay, retention math is secondary to getting paid at all. That path is still how to get your first 10 paying customers without ads.
What to stop doing
- Treating one cancel in a list of fifteen as a trend.
- Copying an enterprise “under 1%” target onto a $29 plan.
- Changing the formula the month the number looks bad.
- Calling a failed card product-market fit.
- Inventing a personal “must hit X% by month six” rule that no public page stated.
- Buying traffic before you know why the last three people left.
When the number finally means something
Trust the rate more when one cancel no longer swings the month by several points, when you have used the same formula for several months, and when you can split new cohorts from older ones. Then the public tables are a real comparison. Until then, the work is people: who stays, who leaves, and whether you sold the right problem.
If you have a real retention story with real numbers and your name on it, send us the story. When the product is real, put it on the map. You can join Indiecity while you fix the leaky month. That gives you people to talk to. It does not change who left.
FAQ
Questions people get stuck on
What is a good monthly churn rate for early SaaS?
There is no single honest target for a product with twenty customers. [ChartMogul’s published median](https://chartmogul.com/saas-metrics/customer-churn/) for companies under $300k ARR is 6.5% monthly customer churn. That is a median across a large anonymized set, not a grade for your first year. With a tiny base, treat it as context, not a weekly score.
Why does my churn look terrible with only 10 or 20 customers?
Because the math is noisy. One cancel out of ten is 10% customer churn that month. Two cancels is 20%. That does not mean your product is doomed or that you are worse than a $3M ARR company with a 3.7% median. Wait for more months and more accounts before you treat the percentage like a board metric. Until then, read the exit reasons.
Customer churn or revenue churn: which one matters?
Both, for different jobs. Customer (logo) churn counts people who leave. Revenue churn counts dollars that leave. If one big account cancels, revenue churn spikes while logo churn looks mild. ChartMogul’s metrics library walks through that split. Early on, know both numbers, then spend your week on the people who left, not on polishing a dashboard.
Is 5% monthly churn fatal?
ChartMogul notes that 5% monthly customer churn compounds to about 46% over a year if nothing changes. That is serious at scale. Early on it can also be a few wrong-fit buyers and a failed card. Do not invent a personal “must hit X% by month six” rule. Fix onboarding, fit, and failed payments first. Re-check the rate when the base is large enough that one leave does not rewrite the story.
Should I aim for under 1% like enterprise SaaS?
Not as a first-year scoreboard. ChartMogul shows lower medians as ARR and price rise. Those companies sell expensive seats with long contracts and dedicated success teams. A solo product at $29 a month will not look like that. Copying the number without that model is pretending you run their company.
When is my sample large enough to trust the percentage?
When a single cancel no longer moves the monthly rate by several points, and when you have several months of the same definition. Until then the percentage is a noisy signal. Your useful data is who left, how long they stayed, whether the card failed, and whether they were ever the buyer you meant to sell to.
Does switching everyone to annual fix churn?
Annual billing can cut quiet monthly exits and pull cash forward. It does not fix a product people never use. People who already paid for a year still leave at renewal if value never showed up. Offer annual when the monthly price already sells. Do not use a yearly lock as a substitute for fit.
What should I track this week if the percent is unreliable?
Count active paying customers at the start of the month and at the end. Note every cancel and every failed payment. Write one line per exit: wrong fit, never activated, price, competitor, card failed, company closed. That list is more useful than a fake benchmark you invented for your niche.
How do I cut churn without a customer success team?
You are the team. Reply the same day when someone is stuck. Watch who never finishes setup. Ask a short cancel reason before you accept the cancel. Fix failed cards with a simple email when the card dies. Stop selling hard to people who were never the buyer. [How to reduce churn with twenty customers](/stories/reduce-saas-churn-20-customers) is that week of calls.
