Segment your churn data: single pattern reveals three different problems
Churn is not a GAAP metric. Every company defines it differently, and most blend numbers that should never sit in the same bucket.
Jason Lemkin learned this early as a B2B CEO. He compared his company to a public competitor that excluded first-60-day churn from their reports. Their reasoning: trial periods do not represent long-term risk. The real reason: churn was much higher in those first 60 days.
That led him to segment churn into three buckets. The patterns were clear:
Single-seat deals under $99/month: 3% monthly churn by revenue. Credit cards expire. Jobs change. Solopreneurs go under. This matches what many single-person businesses see today.
$99 to $999/month deals: 100% net revenue retention after churn. Similar to what HubSpot and Zendesk report in this segment. Small business customers who stick around tend to expand.
$10k to $100k+ annual deals: 120% net revenue retention. Enterprise customers behave like other B2B SaaS buyers. See Box, Salesforce. They stay for years if you solve a real problem.
Churn should be naturally higher in smaller segments. If you do not segment it, no one sees that. You end up trying to solve three different problems with one answer. Keeping enterprise customers happy is not the same as stopping small businesses from switching to a cheaper tool.
Segment NPS and CSAT the same way. Satisfaction varies wildly by customer size. Double down on the happiest, fastest-growing segments.
If you want to exclude trials and POCs from churn, fine. Just do not count them as recurring revenue until they convert. Segment them out as trial revenue. Churn drops when you separate POC revenue from post-trial ARR.
Track where the money actually goes. Even if churn looks high in certain metrics, retaining the revenue is what matters. Blended metrics confuse things. Big customers stay for years. Small customers come and go, even when they are happy.
Segment your data. You will see new patterns. And it might not be as bad as you think. Just different.