Churn rate: measuring it in a way that tells you what to fix

Churn rate is the proportion of customers — or of revenue — lost over a period. It is the most quoted metric in subscription businesses and the most casually calculated: the denominator is often wrong, the period is often unstated, and the choice between counting customers and counting money is usually made by accident rather than deliberately.

The definition itself is simple. Customers lost during the period, divided by customers at the start of the period. What makes it useful is being consistent about the details and looking at more than one version of it.

The versions worth calculating

• Customer churn: customers lost divided by customers at the start. Answers how many relationships you fail to keep.
• Revenue churn: MRR lost divided by MRR at the start. Answers how much it cost you, which is rarely proportional.
• Net revenue churn: revenue churn minus expansion from remaining customers. Can be negative, which is the strongest position a subscription business can be in.
• Voluntary versus involuntary: customers who chose to leave versus those whose payment failed. These have completely different fixes, and involuntary churn is often a quarter of the total.
• By cohort: churn among customers acquired in the same month, tracked over their lifetime. This is where you see whether the product is improving.
• By segment: small customers usually churn far more than large ones, and an aggregate number hides both.
• The period, stated plainly. Monthly and annual churn are different numbers and are constantly confused.

Do not annualise monthly churn by multiplying by twelve. Compounding means 5% monthly churn is roughly 46% annually, not 60%, and the shortcut overstates the loss in a way that produces panic rather than action. If you report annual figures, calculate them properly or state that the monthly figure is monthly.

Measuring it honestly

1. Fix the denominator: customers or MRR at the start of the period, excluding anyone who joined during it. Including new joiners understates churn and is the single most common error.
2. Decide when a customer counts as churned — the day they cancel, or the day their paid period ends — and apply it consistently.
3. Separate voluntary from involuntary, because failed cards are a payments problem rather than a product one.
4. Calculate both customer and revenue churn every period. If revenue churn is much worse, you are losing your larger customers.
5. Break it down by segment and by cohort once you have enough data. Aggregate churn is an average of very different behaviours.
6. Record a reason for every cancellation, in a fixed short list plus free text. This is the input that actually drives fixes.
7. Review the reasons monthly and pick one to address, rather than treating churn as a single problem.
8. Track the trend rather than the level — the absolute number depends heavily on your market, and the direction is what you control.

Involuntary churn is the cheapest to fix

A meaningful share of subscription cancellations are not decisions at all: an expired card, a failed payment, a bank block. Where that is happening, the fixes are mechanical and unusually cost-effective — warn before expiry, retry failed payments on a schedule, tell the customer their payment failed in a way that does not look like marketing, and make updating a card easy. Companies chasing product-driven churn while ignoring this are solving the harder problem first.

Ettex CRM is where the customer side lives: contact profiles with custom fields for plan, start date and status, groups and tags to separate segments and flag at-risk accounts, notes and activity logging so the conversations before a cancellation are visible rather than remembered, a visual pipeline you can use for renewals, and change history showing who did what and when. Cancellation reasons collect cleanly through Ettex Forms with a fixed list plus free text, landing timestamped in one searchable inbox; the churn calculation itself belongs in Ettex Sheets alongside the MRR model; and the invoices that reveal failed payments sit in Ettex Invoices with a status on each one.

Said plainly: Ettex has no subscription or retention tooling. There is no churn calculation, no cohort analysis, no health scoring, no dunning for failed payments, no cancellation flow and no alerts when an account goes quiet. It holds customer records, collects reasons through a form and keeps the invoices; the measurement is a spreadsheet you maintain. Retention at scale is a category of software in its own right. Split it before acting: involuntary churn from failed payments is a dunning management problem, not a product one.

Retention only means something against a revenue base. Monthly recurring revenue is that base: contracted subscription income normalised to a month, which makes growth and loss comparable period to period.

Retention is decided between the renewals, not at them. An account plan records who decides, what the customer is judged on this year, and how many people would notice if you disappeared — which is the number that predicts churn best.

How churn numbers mislead

• New joiners included in the denominator, understating the rate.
• Customer churn quoted when revenue churn is much worse, hiding the loss of large accounts.
• Monthly churn multiplied by twelve, overstating annual loss.
• Voluntary and involuntary mixed, so the cheapest fix is invisible.
• One aggregate figure across segments that behave completely differently.
• Cancellation reasons not recorded, leaving improvement to guesswork.
• Churn compared to an industry benchmark rather than to your own trend.
• The metric watched but never acted on, which is the most common outcome of all.

Churn rate is worth measuring in four versions: customers, revenue, voluntary and involuntary. Get the denominator right, never multiply monthly by twelve, record a reason for every cancellation — and fix the failed payments first.