Churn diagnostics

You can grow through the front door as fast as you want, if the back door is leaking, you're filling a bathtub without a plug. Churn is the second-most-important number in any recurring-revenue business after new bookings. And most teams don't diagnose it well enough to fix it.

The two churn numbers

• Logo churn. % of customers who leave, regardless of size
• Revenue churn. % of recurring revenue that leaves, weighted by contract value

Both matter. A business can have 10% logo churn but 2% revenue churn if only small accounts leave. The opposite is much worse: 2% logo churn but 20% revenue churn means you're losing your whales.

Gross vs, net

• Gross Revenue Retention (GRR), starting ARR minus churn minus downgrades, divided by starting ARR. Max 100%.
• Net Revenue Retention (NRR). GRR + expansion from existing customers. Can exceed 100%.

Benchmarks for SaaS:

• GRR > 90%: healthy. > 95%: best in class.
• NRR > 110%: strong. > 120%: excellent.

If GRR is strong but NRR is weak, your retention is good but expansion is broken. If both are weak, the product-market fit is in question.

Voluntary vs, involuntary churn

• Voluntary, customer chose to leave
• Involuntary, payment failure, credit card declined, company went out of business

Involuntary churn is often 20–40% of total churn and is fixable with dunning processes, card-updater services, and proactive billing outreach. The rest of this page is about voluntary churn.

The five reasons customers leave

1. Didn't realize value. They never got to first value. Onboarding broke. Feature too hard to adopt.
2. Lost value. Their champion left. Their use case changed. Product stopped matching their workflow.
3. Found better alternative. Competitor gave them a better offer or a new capability you don't have.
4. Budget cut. Macro, company-level, or departmental. Not about you.
5. Bad relationship. Support failures, broken promises, account team turnover, outages.

Each requires a different fix. Lumping them together as "churn" hides the root cause.

The churn diagnostic process

For every lost customer (above a threshold, say $10K ARR):

1. Exit interview. Not by their AE, by customer success or a leader. 20 minutes, structured questions.
2. Root cause classification. One of the five reasons above.
3. Timeline reconstruction. When did the seeds of churn get planted? 30 days ago? 6 months ago? Day one?
4. Save attempt (if relevant). Sometimes a concession, a feature commit, or an executive call flips a churn decision.
5. Monthly aggregation. Categorize, count, look for patterns.

Leading indicators, catch it before they leave

By the time a customer says "we're cancelling," you've lost. The fight is upstream. Track:

• Usage, active users, feature adoption, session frequency. Declining usage = risk.
• Health scores, composite metric combining usage, support tickets, NPS, executive sponsor tenure.
• Support ticket sentiment, escalated tickets, frustration tone, repeat issues.
• Sponsor changes, if the person who bought you leaves, your renewal is at risk.
• Business changes, acquisitions, layoffs, reorgs at the customer all raise churn risk.

Customers go through three phases before cancellation: quiet disengagement, then unhappy but still paying, then explicit cancellation. The first two are where you intervene.

Expansion vs, retention

Expansion is usually more valuable than new logo acquisition (lower CAC, higher margin). But expansion and retention are different motions, retention is about preventing leaving; expansion is about deepening usage. Different playbooks. Usually different teams. If your CS team is measured only on retention, expansion will underperform.

Cohort analysis

Aggregate churn hides everything. Always look by cohort (sign-up month/quarter). Questions to answer:

• Are newer cohorts churning faster or slower than older ones?
• Is there a "death month", month 3, 6, 12 where churn spikes?
• Does churn differ by acquisition channel? By segment? By product tier?
• Is post-onboarding churn different from year-2 churn?

What good looks like

• GRR, NRR, and logo churn reported monthly
• Every lost account > $X ARR has a documented root cause
• Health scores trigger proactive outreach before renewal conversations
• Churn root causes feed product roadmap, onboarding fixes, and ICP refinement

Related: Customer success ops · Unit economics · Funnel math