Working out monthly customer churn for a SaaS business
How a SaaS business calculates its monthly customer churn rate, which customers count as churned, and how to read the inputs from invoices and receipts.
By Achekam · · 4 min read · For Software (SaaS) businesses in South Africa, Nigeria, Kenya
Ask three people in a small software company what last month's churn was and you may get three answers. The founder counts cancellations received. The support lead counts accounts that went quiet. The bookkeeper counts customers who stopped paying. Each is measuring something real, and none of them matches the others.
Churn is one of the first numbers an investor asks for and one of the easiest to get inconsistent. The formula is simple. The hard part is deciding, before you calculate, exactly which customers count. This article covers the formula, the decisions behind it, and how to take the inputs from the books rather than from memory.
The formula
Harvard Business Review defines customer churn rate as the percentage of customers who end their relationship with a company in a particular period. It is calculated as the customers who left during the period divided by the total customers at the beginning of the period.
Stripe Atlas defines churn the same way in practice, as the percentage of customers over a given period who do not continue paying, and gives an example: if 200 customers pay in January and only 190 of them pay in February, churn is 5%.
Two details in that example are worth noticing:
- The denominator is the opening count. Customers who joined during February are not part of the calculation; they could not have churned from January's base.
- The test is paying. Stripe Atlas counts customers who did not continue paying, which is a cleaner test than who sent a cancellation e-mail.
Why SaaS measures churn by the month
Harvard Business Review notes that an annual churn rate is the default for most companies, but that businesses which price monthly, such as mobile phone providers, gyms and software-as-a-service companies, track churn by month. A monthly-billed customer can leave every month, so a monthly measure shows problems while there is still time to act. An annual figure would hide a bad quarter inside a good year.
If your customers are mostly on annual contracts, a monthly churn figure is less meaningful, because few contracts come up for renewal in any one month. In that case, measure churn on the renewals due in the period.
Deciding who has churned
Write these rules down before the first calculation, and keep them:
| Situation | Count as churned? | Reasoning to agree |
|---|---|---|
| Customer cancelled and stopped paying | Yes | The clearest case |
| Customer still has access but has not paid for a set period | Decide a cut-off and apply it | Unpaid is not the same as cancelled, but it can be |
| Customer paused their account | Usually no, but track separately | They may come back; watch how many do |
| Customer downgraded to a cheaper plan | No, for customer churn | It is revenue churn, measured separately |
| Customer cancelled and rejoined in the same month | Usually no | Treat as continuous, but note it |
The unpaid row matters most. A customer who stopped paying three months ago but was never formally cancelled will flatter your churn figure if you only count cancellations. Pick a rule, such as "no payment for an invoice more than a set number of days overdue", and apply it every month.
Reading the inputs from the books
The opening count and the leavers can both come from the receivables records rather than from a spreadsheet someone maintains by hand.
The opening count. A business creates customer invoices as drafts and issues them, and issuing takes the next number in the sequence and posts the invoice to the ledger as debtors, sales and output VAT. The number of customers with a recurring invoice issued for the opening month is your starting base.
Who kept paying. A business records customer payments against one invoice or spread across a customer's invoices, and a part payment leaves the invoice part-paid. Compare the customers invoiced at the start of the period with those who paid for the following month, as in Stripe's example.
Who stopped paying. Receivables are aged into 0-30, 31-60, 61-90, 91-120 and over 120 days, with overdue totals. A customer whose invoices are drifting into the older columns is a churn candidate under your unpaid rule, whether or not they have said anything.
No magic number, so split by segment
Harvard Business Review says there is no magic churn number: what is acceptable varies widely by business model, and depends on how quickly and efficiently a company can win customers and how profitable they are. It recommends looking at churn by customer segment, to show which kinds of customers are at risk and may need an intervention.
So instead of hunting for an industry benchmark, split your own churn:
- by plan or price tier;
- by how the customer was acquired;
- by customer size or industry;
- by how long they have been a customer.
A business may find that small customers on the cheapest plan churn quickly while larger ones barely churn at all. That points at a fix, which a single blended number never does. Harvard Business Review's article on keeping the right customers covers what churn measures and which customers are worth keeping. Churn feeds directly into how long your cash lasts, covered in runway and burn, and into what investors expect to see each month, covered in monthly reporting to investors.
Write your churn rules on one page this week, calculate last month's churn from the receivables records using them, and split the result by plan before you share it.
General guidance only, not advice on your own circumstances. Tax and payroll rules change: check the current position with the authority or your adviser.
