Cohort Analysis

RR
Ryan Rutan

Cohort Analysis

Cohort analysis is the practice of grouping users by a shared characteristic and tracking their behavior over time as a group. The shared characteristic is most often acquisition date, but also acquisition channel, plan tier, geography, or onboarding path. It is used to surface trends and inflection points that blended averages hide, and is the standard methodology underneath retention measurement, LTV calculations, and most credible product-market-fit assessments.

The canonical cohort chart is a triangle: rows are cohorts (e.g., users who signed up each week or month), columns are time periods since signup (Day 1, Day 7, Day 30, Month 3, Month 12), and each cell shows what percentage of that cohort was still active or paying at that time. Reading the triangle: look down each column to see whether newer cohorts are retaining better, worse, or the same as older cohorts (the leading indicator of product changes working); look across each row to see how a single cohort decays over time (the cohort curve). A healthy cohort curve flattens at some non-zero percentage, signaling that the product has found a stable core of users who keep coming back. A curve that decays continuously to zero, no matter the cohort, is a leaky bucket. Useful benchmarks: top-quartile consumer apps see Day 30 retention of 25 to 40 percent and flatten by Day 60 to 90. Top SaaS products see annual logo retention above 90 percent and revenue retention above 110 percent. The single most important habit cohort analysis builds is suspicion of single-number metrics: a "60 percent retention rate" can describe wildly different businesses depending on the cohort curve shape.

Ryan's Take

Cohort analysis is the discipline that separates founders who understand their business from founders who are guessing about it. The blended numbers in a top-line dashboard are an average of cohorts with completely different shapes, and the average tells you almost nothing about which cohorts are healthy. The founders who consistently make good product and marketing decisions are the ones who can pull up a cohort triangle in 30 seconds and point at which week things changed. The ones who can't are flying blind and don't know it.

What founders get wrong: Looking at one cohort and declaring victory or defeat. Cohorts are noisy in absolute terms; the signal is in the trend across cohorts. A single cohort with strong retention proves the product can work for someone. Three consecutive cohorts retaining better than the prior ten proves you found a change that works for the next ten.

Related: Retention · Churn Rate · Net Revenue Retention · LTV · Unit Economics

FAQ

What is cohort analysis?
The practice of grouping users or customers by a shared characteristic (usually acquisition date) and tracking their behavior over time as a group, used to surface trends that blended averages hide. The standard methodology underneath retention, LTV, and product-market-fit measurement.

How do you read a cohort retention chart?
Rows are cohorts (e.g., users who signed up each week); columns are time periods since signup (Day 1, 7, 30, etc.). Read down a column to see if newer cohorts retain better than older ones (leading indicator of product changes working). Read across a row to see a single cohort's decay curve.

What does a healthy cohort curve look like?
It flattens at some non-zero percentage, signaling a stable core of users who keep coming back. Top-quartile consumer apps see Day 30 retention of 25-40% and flatten by Day 60-90. A curve that decays continuously to zero is a leaky bucket, regardless of acquisition volume.

Find this article helpful?

This is just a small sample! Register to unlock our in-depth courses, hundreds of video courses, and a library of playbooks and articles to grow your startup fast. Let us Let us show you!

OR

GoogleLinkedInFacebookX/Twitter

Submission confirms agreement to our Terms of Service and Privacy Policy.