# Churn with 20 customers

> With 20 customers, one cancellation is 5% churn. How much the number moves on its own, and how to read it when you're small.

By Luis Miguel, founder of TrustChurn. Published 2026-10-31.
Web version: https://trustchurn.com/blog/churn-with-few-customers

With 20 customers, your monthly churn is mostly noise. One cancellation is 5% churn, two is 10%, and a month where nobody leaves is 0%, even if nothing about the product changed. To read it at that size, measure over 90 days instead of one month, look at the count of customers lost before the percentage, and read each cancellation on its own.

## How much does it move on its own?

Suppose your product's real churn is 3% a month, which is good. With 20 customers you'd expect 0.6 cancellations a month on average. Since you can't lose 0.6 of a customer, what you actually see month to month is roughly:

| Cancellations | Churn that month | How often |
| --- | --- | --- |
| 0 | 0.0% | 55% of months |
| 1 | 5.0% | 33% of months |
| 2 | 10.0% | 10% of months |
| 3 or more | 15.0% or more | 2% of months |

So without anything changing, you'll have a 10% month about once a year. And more than half your months will show 0%, which isn't real either.

React to each month and you'll end up changing prices after a 10% month, or feeling good about a 0% one, when both were mostly chance.

## Does a 90-day window fix it?

It helps, and it doesn't make the number exact. With the same 20 customers and the same 3% real churn, you'd expect 1.8 cancellations over 90 days. Converted to a monthly rate ([how the conversion works](https://trustchurn.com/blog/how-to-calculate-churn-from-stripe)), here's what you'd see:

| Lost in 90 days | Monthly churn | How often |
| --- | --- | --- |
| 0 | 0.0% | 17% |
| 1 | 1.7% | 30% |
| 2 | 3.5% | 27% |
| 3 | 5.3% | 16% |
| 4 | 7.2% | 7% |
| 5 or more | 9.1% or more | 4% |

The steps are smaller (1.7 points per customer instead of 5) and the result lands near the real 3% much more often. But a SaaS with a real churn of 3% will still show anything from 0.0% to 5.3% most of the time, so at 20 customers it's better read as a range.

## When does it start to mean something?

The noise shrinks as the base grows. With 100 customers and the same 3% real churn, you'd expect 3 cancellations a month, and each one moves the number by 1.0 point:

| Cancellations in a month | Churn | How often |
| --- | --- | --- |
| 0 or 1 | 0.0% to 1.0% | 20% |
| 2 to 4 | 2.0% to 4.0% | 61% |
| 5 or more | 5.0% or more | 18% |

That's still a lot of movement for a single month, but the 90-day number at 100 customers is steady enough to compare with last quarter. There's no magic size where churn becomes "real". It gets steadier the more customers and the longer the window.

(All the tables here are worked out with a simple probability model, rounded. They show the size of the swings, not a forecast for any SaaS.)

## How should you read churn when you're small?

**Use a longer window.** Over 90 days you see three months of cancellations instead of one, and the number settles down. That's why TrustChurn measures over 90 days and converts the result to a monthly rate.

**Look at the count.** At 20 customers, "we lost 2 this quarter" tells you more than "3.5% monthly churn". The percentage has a decimal point, but at this size it isn't that precise.

**Read every cancellation.** When you're small you can still email each one and ask what happened, and that tells you more than the count does. Stripe can ask them on the way out ([collecting cancellation reasons](https://trustchurn.com/blog/stripe-cancellation-reasons)).

**Separate the kinds of leaving.** A customer whose card failed didn't decide to go, but once the subscription ends they count as lost all the same. Look at them apart from the real cancellations, because they need a different fix ([failed payments in Stripe](https://trustchurn.com/blog/failed-payments-stripe)).

**Watch what's coming.** Scheduled cancellations and trials ending without a card tell you about next month's losses before they happen ([scheduled cancellations](https://trustchurn.com/blog/scheduled-cancellations-stripe)). With few customers, you can write to each of them personally.

## Should you share your churn at this size?

You can, as long as it's clear what it's based on. "Monthly churn of 1.7%, based on 20 customers over 90 days" is honest and useful. "1.7% churn" alone suggests more precision than there is. If you build in public, you can share it anyway and explain the swings as they come.

## Small SaaS on the ranking

This is also why the TrustChurn ranking has minimums. A SaaS with 4 customers and nobody gone in 90 days has 0.0% churn, and that says very little. To get a place, a SaaS needs a minimum number of paying customers, days of history and customers who could have left in the window.

The minimums aren't fixed: each one is the 25th percentile of the SaaS on the site, with a floor of 3 paying customers and 30 days of history. So while most SaaS on the site are small, a small one can still get a place. The ones that don't meet them yet are listed under "Building history", with what they still need. The [methodology page](https://trustchurn.com/methodology) shows the current minimums.

## FAQ

### How many customers do I need for churn to mean something?

Nobody can give you a cutoff, because the number gets steadier gradually as you add customers and lengthen the window. At 20 customers, a 90-day number still moves by 1.7 points per customer lost. At 100, a single month moves by 1 point per customer, and a 90-day number is steady enough to compare quarter to quarter.

### Why did my churn jump to 10% this month?

With 20 customers, two cancellations in a month is 10%. If your real churn is around 3%, that happens in about one month out of ten without anything changing. Look at who left and why before you decide it's a trend.

### Is 0% churn possible?

For a month or a quarter, yes, especially when you're small. Over a long period with many customers, almost never. With 20 customers and a real churn of 3%, more than half of all months show 0% (see the first table), so a 0% month mostly means nobody happened to leave.

### What should I track instead of churn when I'm small?

The count of customers lost each month, why each one left, and what's coming: scheduled cancellations, failed payments and trials about to end. Those tell you what to do next. The percentage becomes useful once you have more customers.

### Why does TrustChurn use 90 days?

A single month moves too much at this size: in the tables above, one cancellation is 5 points over a month and 1.7 over 90 days. A window much longer than a quarter would bury recent changes, like a new price. TrustChurn converts the 90-day result to a monthly rate so it still compares with other monthly numbers.

---

Keep reading:

- [How to calculate your monthly churn from Stripe](https://trustchurn.com/blog/how-to-calculate-churn-from-stripe.md)
- [How to ask customers why they cancel, in Stripe](https://trustchurn.com/blog/stripe-cancellation-reasons.md)

TrustChurn is a public ranking of SaaS by monthly churn, read from Stripe with a read-only restricted key, not typed by the founder. The formulas are public: https://trustchurn.com/methodology