Aggregate churn numbers hide the real story. Cohort analysis by signup month reveals whether retention is improving.
Aggregate monthly churn — 'we churned 3% last month' — is one of the most misleading metrics in SaaS. It blends new customers still in the honeymoon period with customers who've been around long enough to reveal their true stickiness. It masks whether retention is improving or degrading over time. And it makes it nearly impossible to attribute changes in churn to specific product or GTM decisions. Cohort analysis — grouping customers by when they signed up and tracking each cohort's retention curve over time — is the fix. It's not new or clever, but it's chronically underused by early-stage teams who don't yet have a rhythm for looking at it.
Group customers by signup month (or week, for high-velocity businesses). For each cohort, plot % of original cohort still active at month 1, month 2, month 3, and so on. You'll get a set of curves, one per cohort. Log-scale the y-axis if curves are all above 90%; linear scale if below. Do this for logo retention (are they still customers?) and for dollar retention (how much revenue is that cohort producing vs. what they started at, capturing both churn and expansion). Both matter, and they tell different stories.
Curve that drops fast in months 1-3, then flattens: onboarding problem. Customers who make it past the initial 90 days stick around, but many don't make it. Fix: onboarding, activation criteria, early-tenure success programs. Curve that declines steadily forever: no natural retention floor. The product isn't sticky enough that users become permanent. Fix: usually product depth, workflow integration, or switching-cost creation. Curve that plateaus at 70-80%+ after month 6: healthy sticky product with real retention. Focus on expansion rather than churn reduction. Curves getting flatter over cohort vintages: retention improving over time (product-market fit improving). Curves getting steeper: something is regressing — often quality, competition, or a bad recent cohort acquisition motion.
Dollar retention (aka net dollar retention or NDR) treats each cohort's initial ARR as 100% and tracks what that cohort produces over time. A cohort that lost some logos but grew its remaining logos aggressively can show dollar retention above 100% — the classic 'best of SaaS' signature. Dollar retention curves that stay above 100% mean the business compounds even without new customer acquisition. Curves that decline mean the business must acquire faster than it churns just to stay flat.
Slice cohorts by: acquisition channel (paid vs. organic vs. sales-led — often dramatically different retention), initial deal size (small SMB deals often churn 3-5x faster than mid-market), industry vertical, plan tier, product used first, geography. Nearly every business finds a segment that's meaningfully better or worse than the aggregate. That insight is directly actionable — double down on the good segments, fix or exit the bad ones. Aggregate churn number tells you none of this.
Review cohort retention monthly at leadership level, weekly at CS/product level during active retention work. Tools: Amplitude, Mixpanel, Hex, ChartMogul, ProfitWell, or a bespoke SQL + BI dashboard. The tooling isn't the constraint — the constraint is committing to look at the curves, discussing what changed, and connecting changes to specific product or GTM decisions. Companies that make cohort review a monthly ritual improve retention 2-3x faster than those who only look during board prep.
Investor directory · Fundraising library · Articles A–Z · Company funding database