How do I calculate churn when new customers keep arriving?
Divide the customers who left during the period by the customers you had at its start, and leave new arrivals out. Then do the same with recurring fees. In the example one of ten customers left (10%), but they paid €500 of €2,000 a month (25%). A newcomer restores the count, not the loss.
A new customer can replace someone who left without replacing their revenue. This example shows why the final customer count can hide a loss, and why a missing repeat order needs a different check.
Ask who remained from the original group
Customer churn measures a defined loss over a stated period. For the starting-group measure used here, divide departures from that group by its opening customer count. New arrivals belong in a separate count; they cannot restore a departed member of the original group.
Keep the population consistent. Someone who joined and left within the month was not in its starting group. Report that event separately or state a different measurement rule. Cancelling early also does not, by itself, tell you why the customer left. Timing is an observation; the reason needs evidence.
Ten customers before and after, but €300 less in monthly fees
A fictional maintenance business opens June with ten contracted customers paying €2,000 in total monthly fees. During June, one customer paying €500 cancels. The other nine remain on unchanged terms.
The business has lost one of its original ten customers: 10%. It has lost €500 of the original €2,000 monthly fees: 25%. The account was only one customer, but a much larger share of the money.
A new customer then joins at €200 a month. By 30 June there are ten customers again, with monthly fees of €2,000 − €500 + €200 = €1,700. Nine of the original ten stayed. The newcomer replaced the headcount, but did not undo the departure or replace all its fees.
This is what the example explains: follow the original customers and the fee amounts separately. An unchanged final count does not mean zero loss. These are monthly fee amounts; the example does not calculate June's cash receipts or profit.
- This is a fictional contract business measured from the start of June to 30 June.
- There are no other cancellations, price changes, downgrades, upgrades or reactivations.
- Monthly fees exclude tax. Billing dates, proration, receipts and delivery costs are outside this simplified comparison.
| Group or event | Customers | Monthly fees |
|---|---|---|
| Opening group | 10 | €2,000 |
| Departure from that group | 1 lost out of 10 = 10% | €500 lost out of €2,000 = 25% |
| Original customers remaining | 9 | €1,500 |
| Separate new arrival | 1 | €200 |
| Total at 30 June | 10 | €1,700 |
Try it with your own numbers
The group you follow through the period.
Confirmed departures only. A missing order may only be late.
Counted separately. They do not undo a departure.
What the starting group paid each month, without tax.
Fees lost through cancellations or downgrades in the starting group.
What the new arrivals pay each month.
Your result
Fill in every field to see the result.
Your numbers stay in this browser. Nothing you type is sent or saved.
A lost account and lost monthly fees have different weights
Customer-count loss gives each customer one place in the count. Gross recurring-revenue loss compares recurring fees lost through cancellations or downgrades with opening recurring fees. Net revenue churn also accounts for expansion from existing customers. These measures answer different questions.
In the maintenance example, the departing customer is one of ten but pays one quarter of the opening monthly fees. The 10% and 25% results therefore fit together. Neither is lost profit or cash missing from the bank. For project work, a departed buyer's past sales describe historical exposure; they are not automatically recurring fees lost this month.
Silence may mean a pause
Consider a separate fictional exhibition buyer who usually orders before an annual event. This year's usual order date passes without a purchase. The event may have been postponed. The missing order is a reason to check, not proof the relationship has ended.
For irregular purchasing, state what you mean by inactive and why that rule fits the usual buying pattern. Keep confirmed departure, suspected inactivity and an unknown reason separate. A one-time buyer who was never expected to return should not silently enter the same loss count as a cancelled subscription.
Warning signs also differ from actual departures. Fewer accounts marked at risk could mean improvement, but could also mean those accounts left or disappeared from the list. A completed check-in records an action. It does not, on its own, prove that the action retained a customer.
Common misconceptions
- The same final headcount means everyone stayed.
- New arrivals can conceal departures in the total. Compare the identities in the starting group, not just two totals.
- No purchase this month means the customer was lost.
- Check the normal buying pattern and what actually happened. Inactivity and a confirmed cancellation are different evidence.
- The customer left quickly, so onboarding caused it.
- A date establishes when the departure happened. It does not establish the cause. Preserve an unknown reason until there is evidence.
Check one departure and one quiet relationship
For each, record the relevant dates, normal purchase or renewal pattern, evidence of departure and known reason. Leave uncertain reasons marked unknown. For a loss rate, retain the matching starting group and record new arrivals separately. Compare the customer count and any recurring-fee loss under their own labels before choosing what to investigate.
Sources
- Stripe: How to calculate churn rates (opens in a new tab)A stated period and opening customer count for measuring customer loss. This example explicitly restricts departures to the same opening group.
- Stripe: Revenue churn (opens in a new tab)Recurring-revenue loss through cancellations or downgrades, and the difference between gross loss and a net measure including existing-customer expansion.