A cohort is a group of users sharing a defined start event — for example, customers acquired in January 2025 — tracked over time to measure behavior. Cohort analysis is central to unit economics because cohorts isolate acquisition-channel quality and product-market fit; blended metrics hide deteriorating economics masked by newer, better cohorts. The classic retention curve shape is a steep drop in the first weeks followed by a long, flat tail, and the goal is to make that long tail as high as possible. A variant known as the "smile" curve dips then rises — typical of products with a learning curve where power users become more engaged over time. DAU/MAU, the ratio of Daily Active Users to Monthly Active Users, measures habit strength, with companies like Facebook famously targeting above 50%. Retention and churn are mirror images: retention is the percentage of customers who stay over a period, churn is the percentage who leave, and the two sum to 100% within a closed cohort.
Churn comes in several flavors, and tracking the right one is essential. Logo churn is the percentage of customer accounts that cancel in a period, regardless of contract size — a measure of customer count loss. Revenue churn is the percentage of recurring revenue lost in a period, including downgrades — a measure of dollar loss. The two can diverge wildly: losing small customers is fine, losing whales is fatal. Gross churn is the total revenue lost to cancellations and downgrades; net churn subtracts expansion revenue from the same cohort, often producing a negative net churn. A "negative churn" SaaS business is one where expansion MRR from existing customers exceeds churned MRR every month, so the existing book grows even with zero new sales — a hallmark of best-in-class SaaS. A "leaky bucket" describes a business with high churn such that new sales barely outpace losses — the bucket never fills regardless of how fast you pour.
Net Revenue Retention (NRR) operationalizes this: starting ARR from a cohort plus expansion minus churn and contraction, divided by starting ARR, expressed as a percentage. Top-quartile SaaS companies achieve above 120%, with some reaching 130–140%+, meaning existing customers grow even as some churn. Gross Revenue Retention (GRR) is the same calculation excluding upsell — the "floor" of retention health — with above 85% annual acceptable for SMB SaaS, above 90% good, and above 95% the enterprise target. Expansion revenue is additional recurring revenue from existing customers via upsells, cross-sells, seat growth, or usage increases; contraction revenue is the mirror image, lost when customers downgrade, reduce seats, or lower usage tier. Net Promoter Score (NPS), which measures willingness to recommend on a 0–100 scale, correlates with low churn at the aggregate level, though the correlation is noisy at the segment level.
Activation sits between signup and retention as the first moment a new user experiences the core value of the product. Activation is critical to unit economics because users who never activate have near-zero LTV, so increasing the activation rate — the percentage of signups who reach the activation event within a defined window — directly increases effective LTV without spending more on acquisition. Activation differs from conversion: activation is the user experiencing product value (often on a free tier), while conversion is the user paying, and both are required for revenue LTV. Other leading indicators of churn include usage drops, support tickets, NPS decline, payment failures, and missing milestones — lagging indicators like formal cancellation arrive too late to act. Voluntary churn is the customer's choice to leave; involuntary churn comes from payment failure (expired cards, insufficient funds) — sometimes called credit card churn — and can be recovered through dunning, the process of retrying failed payments and contacting customers to fix billing, which can reclaim 30–50% of involuntary churn. A "zombie" customer is still on the books but with near-zero engagement and high churn risk, a drag on support cost and an LTV depressor; a "negative customer" is one whose fully-loaded cost to serve exceeds their revenue, deepening losses with every additional month they stay.