Churn: Finding It, Measuring It, and Fixing It Before It Compounds

Churn does not merely reduce growth; it sets a hard limit on how large the company can ever get. That limit has a formula, and it is the most sobering piece of arithmetic in a subscription business.

The ceiling

At a steady acquisition rate, a company stabilises where new customers equal churned customers:

Maximum customers = monthly new customers / monthly churn rate

| Monthly new customers | Monthly churn | Ceiling | | --- | --- | --- | | 100 | 5% | 2,000 customers | | 100 | 3% | 3,333 customers | | 100 | 1.5% | 6,667 customers |

Same acquisition effort, three completely different companies. Halving churn does more for long-run scale than doubling the marketing budget — and it is almost always cheaper.

Measure three things, not one

Logo churn — customers lost divided by customers at period start. Tells you about product fit and onboarding.

Gross revenue churn — revenue lost from cancellations and downgrades, divided by starting revenue. Never negative. Tells you how much value is leaking.

Net revenue retention (NRR) — starting revenue, minus churn and downgrades, plus expansion, divided by starting revenue. Above 100% means existing customers grow faster than they leave.

Worked month:

| Line | Amount | | --- | --- | | Starting MRR | $400,000 | | Lost to cancellations | -$14,000 | | Lost to downgrades | -$6,000 | | Gained from upgrades | $32,000 | | Ending MRR (existing customers) | $412,000 | | Gross revenue churn | 5.0% | | Net revenue retention | 103% |

A 5% gross churn with 103% NRR is a genuinely mixed picture: the base is leaking badly while a few large accounts mask it. Reporting only NRR would hide the problem for a year.

Rough benchmarks

| Segment | Healthy monthly logo churn | Healthy NRR | | --- | --- | --- | | SMB self-serve | 3-5% | 90-100% | | Mid-market | 1-2% | 100-115% | | Enterprise | Under 1% | 110-130% |

Find where it happens

Cohort the churn by tenure. The shape tells you what is broken.

| Churn concentrated in | Real problem | Where to fix it | | --- | --- | --- | | First 30 days | Onboarding or wrong-fit customers | Activation flow, sales qualification | | Months 2-4 | Value never materialised | Time to first value, success outreach | | Month 12-13 | Renewal and budget review | Proving ROI before renewal season | | Spread evenly | Weak product-market fit | Product and segment, not tactics |

Then segment by acquisition channel, plan size and use case. It is common to find that one channel or one plan produces most of the churn — which turns an abstract retention problem into a specific and fixable decision.

The interventions that work, in order

1. Fix activation. Define the single action that correlates with retention — the "aha" moment — and measure what share of new customers reach it in week one. Moving that percentage is usually the highest-return retention work available.
2. Qualify harder at the top. Some churn is sold, not caused. Customers who were never a fit will leave regardless of how good the product becomes.
3. Build an early warning signal. Declining logins, a departed champion, or an unopened report are all leading indicators. Acting a month before renewal beats reacting after cancellation.
4. Make the cancellation flow informative. Not obstructive — informative. A required reason field, plus a pause option and a downgrade option, recovers a meaningful share and tells you why the rest left.
5. Offer annual plans. An annual contract removes eleven monthly opportunities to reconsider. A 15-20% discount for annual prepay usually pays for itself in retention and cash flow alike.
6. Expand deliberately. Seat growth, usage tiers and add-ons are what push NRR above 100%, and they are cheaper to sell than new logos.

Common mistakes

• Reporting only NRR. It hides a leaking base beneath a few expanding accounts.
• Calculating churn on a growing base. Divide by the starting count, not the ending one, or you will systematically understate it.
• Treating churn as a customer success problem. Most of it is decided by product and by who you sold to.
• Discounting to save an account. It resets the price permanently and teaches customers to threaten to leave.
• Ignoring involuntary churn. Failed payments are often 20-40% of total churn and are fixed with retry logic and card-expiry notices — the cheapest retention work there is.

How the simulator models it

Garage to IPO models customer retention as a direct input to revenue compounding. Neglect product quality while pushing growth and the churn rate rises, capping the business well below where the spend implies it should be. The players who reach the public stage with durable valuations are the ones who fixed retention before scaling acquisition.

Building a simple retention dashboard

You do not need a specialist tool to manage churn. Five views, reviewed monthly, cover almost everything that matters.

1. Cohort retention table. Rows are the month customers joined, columns are months since joining. The shape of the curve — whether it flattens and where — is the single most informative chart in a subscription business.
2. Churn reasons, categorised. Group free-text cancellation reasons into five or six buckets. Do not add more categories; granularity here produces noise, not insight.
3. Churn by acquisition channel. Frequently the fastest route to a large improvement, because it turns a retention problem into a marketing budget decision.
4. Activation rate by week. The share of new customers reaching the defined value moment within seven days, tracked as a trend.
5. At-risk list. Accounts whose usage dropped more than a set percentage month over month, reviewed by a named person every week.

Assign one owner to the dashboard and put it in the same monthly meeting as the revenue numbers. Retention work fails most often not because teams cannot fix churn, but because nobody is accountable for looking at it until the quarter has already been lost.

Where to go next

Churn is the largest input into lifetime value — see CAC, LTV and payback — and the fastest way to extend runway without cutting, covered in runway planning and when to cut.