How to Read a Cohort Retention Chart

A cohort retention chart is the single most useful visualization in SaaS analytics. It looks intimidating at first (a grid of percentages, decreasing left to right) but the insights are simple once you know what to look for.

What each part of the chart means

• Rows are cohorts. Usually customers who signed up in a given month. "January 2026 cohort" = everyone who signed up in January.
• Columns are time since signup. Month 0 = signup month. Month 1 = 30 days later. Month 12 = one year in.
• Cell values are retention percentages. "72%" in the [March, Month 3] cell means 72% of the March cohort was still active 3 months after signing up.

Reading the chart:

• Down a column shows how retention at a specific tenure has changed across cohorts. Are newer cohorts retaining better?
• Across a row shows how a single cohort has aged. What percentage survived to each milestone?

The 4 things every chart tells you

1. Where in the lifecycle churn happens

Look at the shape of any row. If most of the drop is in the first 1-3 months, you have an activation problem: customers signed up but never reached the aha moment. If the drop is spread evenly across months, you have an engagement problem: customers activated but drifted over time. If the drop happens around specific renewal points (month 12, month 24), you have a pricing/decision problem.

Different diagnoses need different fixes. See aha moment guide for activation, retention emails for engagement, and annual plans for renewal-window churn.

2. Are newer cohorts retaining better or worse than older ones?

Read down a specific column (say, Month 3). Compare the Month 3 retention of your January cohort to your June cohort. If June is higher, retention is improving. If June is lower, something has gotten harder (worse acquisition targeting, price change, competitive pressure).

This is the earliest indicator of a future churn problem. It shows up in cohort data months before it shows up in blended monthly churn rate.

3. Outlier cohorts worth investigating

Sometimes a single cohort will look dramatically different from its neighbors. If April retention curves are 15% lower than March and May, something specific happened to April. It could be a product bug that affected onboarding, a promotional campaign that brought in wrong-fit customers, or an external event (a competitor launch, a news cycle).

Cohort charts surface these outliers automatically. Blended monthly churn hides them.

4. Your long-term retention floor

Read a row all the way out to month 12, 18, or 24. Does retention stabilize at 60%? 40%? 20%? That is your business's long-term retention floor. Everything else being equal, that is the percentage of customers you keep forever.

The floor is what compounds. Improving your Month-24 retention from 40% to 60% is worth vastly more than reducing Month-1 churn from 15% to 10%.

The 3 patterns that predict SaaS outcomes

Pattern 1: Steep drop, then flat

The healthy pattern. Some early churn is expected (users who signed up to try, decided not for them). Then retention stabilizes and the flat tail compounds. Best-in-class SaaS shows this shape.

Pattern 2: Steady slope toward zero

Bad. You will lose 100% of every cohort eventually. Indicates a value-decline problem: users activate but drift over time. The fix is engagement work: sticky features, expansion revenue, deeper integrations.

Pattern 3: Smile curve (drops, flattens, then rises)

Excellent. Retention actually increases at some point in the lifecycle. Happens when expansion revenue outpaces churn, or when network effects deepen over time. Slack, Airbnb, and Notion show smile curves in their long-term cohorts.

How to build one

Tools that build cohort retention charts automatically:

• Revenue cohorts: ChartMogul, Baremetrics (both Stripe-native)
• Behavioral cohorts: Amplitude, Mixpanel, PostHog
• DIY: Spreadsheet with COUNTIF formulas, or SQL in your data warehouse

See the cohort analysis guide for the full setup and retention curve simulator to project what your data implies for the future.

Score your measurement setup

Take the 60-second Churn Health Check. It scores your measurement maturity (including whether you use cohort analysis) and tells you what to fix next.