Retention and churn

Retention is the foundation that everything else sits on. No amount of acquisition can outrun bad retention — if you’re losing 5% of customers monthly, you’re replacing nearly half your customer base every year. Fix retention first, then scale acquisition.

Customer retention rate

Example

• Start of month: 100 customers
• End of month: 105 customers
• New customers acquired: 10

Retention rate = (105 − 10) ÷ 100 = 95%

Small differences in monthly retention compound dramatically. The table below shows what monthly rates look like over a year:

Going from 5% to 2% monthly churn means keeping 78 customers instead of 54 out of every 100 after one year. That’s a massive difference in LTV, revenue compounding, and growth sustainability.

Churn rate

Churn is the inverse of retention, but it’s worth tracking separately because the breakdown reveals where to focus.

Types of churn

Voluntary churn — customers actively cancel. This is a product, pricing, or competitive problem.

Involuntary churn — failed payments, expired cards. This is an operational problem with a clear fix: dunning emails, card update reminders, retry logic.

Silent churn — customers stop using the product but don’t cancel (especially on annual contracts). Track usage-based churn alongside subscription-based churn to catch this earlier.

When customers churn matters

Early churn (first 90 days) is almost always an onboarding or activation problem. If most of your churn happens here, invest in time to value before anything else.

Net Revenue Retention (NRR)

NRR is one of the most important metrics in SaaS — it tells you whether your existing customers generate more or less revenue over time, accounting for expansion, contraction, and churn.

Example

• Starting MRR: $100K
• Expansion: +$15K
• Churned: −$8K
• Contraction: −$2K

NRR = ($100K + $15K − $8K − $2K) ÷ $100K = 105%

NRR above 100% means you can grow revenue without acquiring a single new customer. This is the most powerful growth dynamic in SaaS — revenue from existing customers compounds month over month, reducing dependence on expensive acquisition.

If your NRR is below 100%, stop investing in new acquisition until you fix the leak. You’re pouring water into a bucket with a hole.

Gross Revenue Retention (GRR)

GRR isolates how well you retain revenue without counting expansion. It’s a purer measure of product stickiness. A company might mask a 92% GRR (mediocre) with strong expansion to show 115% NRR (great) — but the underlying retention problem will eventually catch up.

Cohort retention analysis

Aggregate retention metrics hide trends. Cohort analysis groups customers by signup month and tracks their behavior over time — it’s the only way to know if your product is actually getting better.

What to look for

• Improving cohorts — newer cohorts retain better than older ones, meaning product/onboarding improvements are working
• Retention curve flattening — most SaaS products see initial churn that stabilizes around month 3–6. Where your curve flattens determines long-term value
• Seasonal patterns — customers who sign up in January might retain differently than those in July
• Channel differences — organic signups typically retain better than paid, and sales-assisted often retain better than self-serve

Track both customer cohorts (headcount) and revenue cohorts (dollars retained). Revenue cohorts weight high-value accounts appropriately.

Leading indicators of churn

By the time a customer cancels, it’s usually too late. Build early warning systems using behavioral signals that predict churn 30–60 days before it happens.

Product usage signals:

• Declining login frequency vs their established pattern
• Reduced feature usage (especially core features)
• Decreasing session duration
• Abandoned workflows (starting but not completing key tasks)

Account-level signals:

• Champion leaves the company
• Declining number of active seats vs total seats
• Increasing support tickets or negative sentiment
• Missed business reviews or ignored outreach

Health scoring

Combine multiple signals into a composite health score:

Single indicators are unreliable — a user might skip a week for vacation. Combine 3–5 signals and weight them by how strongly they correlate with actual churn in your historical data.