Why Cohort Analysis Beats Blended Averages
A blended churn rate is a single number that hides everything important about your retention. It tells you that, on average, you lose some percentage of customers each month — but it conceals whether the loss is evenly distributed or concentrated in the first 30 days after signup. Cohort analysis breaks that average apart, grouping customers by when they signed up and tracking each group over time. The resulting picture is dramatically more informative.
Investors request cohort tables during due diligence for a reason: the shape of the retention curve reveals whether you have product-market fit far more reliably than any growth-rate headline. A flattening cohort curve — where retention stops declining after a few months and plateaus — is the signature of a product that has found its audience. A curve that continues to decay toward zero says the opposite, regardless of how fast you are acquiring new users.
A growth rate tells you how fast you are filling the bucket. A cohort curve tells you whether the bucket has a hole in it. The first number is exciting; the second is the one that determines whether you have a business.
Building a Cohort Retention Table
A cohort retention table groups customers by their signup month (the cohort) and tracks what percentage of each cohort is still active in subsequent months. Each row is a cohort; each column is a month offset from signup. The cell at row "January cohort," column "Month 3" shows what percentage of customers who signed up in January were still active in April.
The construction requires only two data points per customer: their signup date and their cancellation date (or last active date for usage-based retention). Assign each customer to a cohort by truncating their signup date to the month. For each cohort, compute the size at month zero (100 percent), then the surviving percentage at month one, month two, and so on. The result is a triangular matrix because older cohorts have more observed months.
What to Measure
Retention can be measured by customer count (logo retention) or by revenue (revenue retention). Both matter. Logo retention tells you whether customers find the product valuable enough to stay. Revenue retention tells you whether the customers who stay are worth more over time. A product with strong logo retention but weak revenue retention may have a pricing problem; a product with weak logo retention but strong revenue retention may be depending on a few large accounts.
Reading the Shape of the Curve
The most important diagnostic from a cohort table is the shape of the retention curve — specifically, whether it flattens. Four archetypal curves appear repeatedly in SaaS businesses, and each implies a different diagnosis and prescription.
- **The flattening curve (good).** Retention drops sharply in the first month, then plateaus at a stable level. The early drop is customers who never found value; the plateau is the customers who did. This is the signature of product-market fit within a segment.
- **The linear decay curve (bad).** Retention declines at a roughly constant rate indefinitely. There is no stable core of loyal customers. This indicates the product delivers one-time value but no ongoing reason to stay.
- **The accelerating decay curve (very bad).** Retention drops faster over time, not slower. Customers leave more readily the longer they have been with you. This often indicates a degradation in product experience, a scaling failure, or a competitor eating your base.
- **The smile curve (rare, powerful).** Retention dips, then recovers as customers who stayed begin to expand. Seen in products with long time-to-value where early churn is high but surviving customers become deeply embedded.
Compare curves across cohorts over time. If each successive cohort retains better than the one before it, your onboarding or targeting is improving. If newer cohorts retain worse, something in your acquisition or product has degraded — often a sign that you are scaling into a less suitable customer segment.
Segmenting Cohorts for Signal
A single cohort table for your entire customer base can average away the most important insights. Segment your cohorts by the dimensions that matter: acquisition channel, plan tier, company size, and use case. A cohort of customers acquired through organic search may retain at 90 percent month-one while paid-acquisition cohorts retain at 60 percent — a signal that your paid channels are bringing in the wrong customers, not that your product is broken.
Plan-tier segmentation is equally revealing. Enterprise cohorts that retain at 98 percent alongside self-serve cohorts that retain at 70 percent tell you where your product-market fit is strongest and where your retention investment should focus. A blended number of 85 percent hides both extremes and leads to generic "improve onboarding" initiatives that help neither segment.
From Diagnosis to Action
Cohort analysis is diagnostic, not prescriptive, but it points directly at the right interventions. A steep first-month drop across all cohorts points at onboarding and time-to-first-value — the customer never experienced the benefit. A flattening curve that plateaus too low points at a segment mismatch — the product works for a narrow audience and you are over-acquiring outside it. A declining curve among newer cohorts points at a recent change: a pricing update, a product regression, or a shift in acquisition mix.
The action is always to investigate the specific cohort and the specific drop-off, not to launch a generic retention program. Cohort analysis gives you the resolution to see exactly where customers are leaving and why. Use the SaaS Churn calculator to quantify the aggregate impact, then build the cohort table to find the segment-level lever that moves it.