Churn Rate Analysis: How to Find the Revenue You Are Losing

Most companies report a churn rate. Far fewer analyse it. A single monthly percentage tells you that customers are leaving. It does not tell you which customers, when in their life they leave, what they were paying, or why. Churn rate analysis is the work of turning one number into a diagnosis you can act on.

This guide covers the formulas that matter, how to read them together, and how to separate the three root causes that produce almost all churn: the wrong customer, a product gap, or a price that does not match value.

The formulas, and what each one hides

Customer churn rate

Customer churn = customers lost in the period / customers at the start of the period

Counts logos. Treats a $50 account and a $50,000 account as the same event, so it overstates the damage in a self-serve base and understates it in enterprise.

Gross revenue churn

Gross revenue churn = (MRR lost from cancellations and downgrades) / MRR at the start

Counts money. This is the number to lead with, because it reflects what actually left the business.

Net revenue retention

NRR = (starting MRR + expansion - contraction - churn) / starting MRR

The most important single figure in a subscription business. Above 100 percent means the existing base grows without new sales. Between 100 and 110 percent is acceptable; 120 percent or higher is where compounding takes over. NRR is where pricing structure shows up most clearly, because expansion is a function of your value metric.

Two rules people get wrong

• Do not annualise a monthly rate by multiplying by twelve. Use 1 - (1 - monthly rate)^12. Three percent monthly is roughly 31 percent annual, not 36.
• Do not mix contract lengths in one bucket. Annual contracts can only churn on their renewal date, so monthly churn for that group is meaningless. Analyse renewal cohorts separately.

Cohort analysis: where churn actually becomes readable

Group customers by the month they signed up and track what percentage remain in month 1, 3, 6, 12. Read the shape of the curve, not the endpoint.

• Steep early drop, then flat. Normal and healthy. You have an onboarding and qualification problem, not a product problem. The flattening point is your real retained base.
• Steady constant decline that never flattens. The serious one. It means the product never becomes indispensable and there is no natural retained core. Growth here is a treadmill.
• A cliff at month 12 or 13. An annual renewal problem. Someone is reassessing value against price and deciding no. Almost always a pricing and value-proof issue rather than a feature gap.
• A curve that bends up. Expansion inside the cohort is outrunning losses. Whatever produced that, do more of it.

Then cut cohorts by dimension: acquisition channel, plan, segment, company size, onboarding path, feature adoption in the first 30 days. Churn is almost never uniform. It concentrates, and the concentration is your answer.

Separating the three root causes

Cause 1: you acquired the wrong customer

Signals: churn concentrated in one channel or one segment, high churn in the first 60 days, small deal sizes, heavy support load before cancelling. The fix is upstream in qualification, positioning and entry price, not in retention campaigns. Price is a filter here, and pricing too low is the most common way companies buy their own churn.

Cause 2: a product or activation gap

Signals: churn correlates with never reaching a specific action or usage threshold; retained accounts share a behaviour that churned accounts skipped. Find the activation event that separates the two groups and rebuild onboarding around it.

Cause 3: price does not match perceived value

This is the cause most often misdiagnosed as either of the others. Signals:

• Churn clusters at renewal dates and after price changes.
• Cancellation reasons mention budget, justification, or "not using it enough".
• Customers downgrade rather than leave, and contraction is a large share of revenue churn.
• Heavy users churn as often as light users, which means your price is disconnected from the value delivered.
• Customers deliberately limit usage or seats to control cost.

That last pattern is the clearest structural signal there is. If a customer suppresses adoption to manage their bill, your value metric is fighting your retention. A seat model in a product where an agent does the work produces exactly this: value rises, seats do not, and the renewal conversation becomes a negotiation about a number nobody can justify.

A churn analysis you can run in a week

1. Establish the baseline. Customer churn, gross revenue churn, NRR, monthly and annualised, for the last 12 months.
2. Build sign-up cohorts and plot the retention curves. Identify the shape and the flattening point.
3. Cut by segment, channel and plan. Find the two or three cohorts carrying most of the loss.
4. Split voluntary from involuntary. Failed payments are often 20 to 40 percent of reported churn in self-serve businesses and are fixed with dunning and card-update flows, not strategy.
5. Read the money, not just the logos. Separate full cancellations from downgrades. Contraction usually points at packaging.
6. Interview twelve churned accounts. Ask what changed, not why they left. People rationalise; timelines do not.
7. Attribute each loss to wrong customer, product gap, or price and value mismatch. Then fix the largest bucket.

Fixing the pricing bucket

When the diagnosis is price and value mismatch, retention tactics will not hold. The structure has to change.

• Move the price onto a metric that grows with delivered value, so the account expands as it succeeds instead of renegotiating. Start with the pricing models comparison.
• Rebuild tiers on evidence. Use Van Westendorp to find the acceptable range and Gabor-Granger to test specific points before you ship them.
• Consider outcome-based structures where the result is measurable, so the renewal argument is made by the data rather than by your account manager. See the outcome-based pricing guide.
• Add guardrails to consumption pricing. Caps, alerts and committed floors remove the bill shock that drives cancellations, as covered in the usage trap.

Churn also feeds straight into acquisition maths: it is the denominator of lifetime value, so a point of churn changes what you can afford to spend to grow. See customer acquisition and pricing.

The short version

One churn percentage is a smoke alarm. Cohorts, revenue churn and NRR tell you which room is burning. And when the answer is price rather than product, the fix is the model, not the message.