Cohort Analysis

Analytics

Also: Cohort · Cohort Report

Retention Rate (Week N) = (Users from cohort active in Week N ÷ Cohort size) × 100

What it isGrouping users by when they joined, then tracking their behaviour over time

Key insightShows whether retention is improving or declining across acquisition periods

Best useDiagnosing churn, evaluating onboarding changes, measuring product improvements

Quick definition

Cohort analysis groups users by a shared characteristic (usually acquisition date) and tracks their behaviour over time, revealing how retention, engagement or revenue evolves across different user groups.

Where it shows up in the data

See Retention benchmarks by industry →

Acquisition cohort

Users grouped by when they first started using the product or made their first purchase. The most common cohort type.

Retention curve

A graph showing what percentage of each cohort is still active at each time interval. Healthy products show the curve flattening rather than declining to zero.

Cohort comparison

Comparing retention curves across different cohorts to see whether product or experience changes are improving retention over time.

Revenue cohorts

Tracking revenue per cohort over time. Often shows negative churn: cohorts that generate more revenue in month 6 than month 1 due to expansion revenue.

What it actually means

Cohort analysis slices your user or customer base by when they joined and follows each group forward in time. Instead of asking 'what percentage of users are active this month?' (which mixes old and new users together), cohort analysis asks 'of the users who joined in March, what percentage are still active in June?'

This distinction is critical. A growing business can have flat or declining retention hidden by the volume of new acquisitions. Cohort analysis surfaces this. It also shows whether changes to onboarding, product or pricing are improving retention for newer cohorts relative to older ones.

GA4's cohort report, Mixpanel, Amplitude and Braze all offer cohort analysis. For e-commerce, it's often easier to run cohort queries directly against order data.

Aggregate metrics hide what cohort analysis reveals: are the people you acquired 6 months ago still here?

How to calculate it

Retention Rate at Week N = (Users from cohort still active in Week N ÷ Original cohort size) × 100

Worked example. 100 users acquired in January. In Week 4: 55 are still active (55% retention). In Week 8: 38 are still active (38% retention). In Week 12: 31 are still active (31% retention). The curve is flattening, suggesting a stable retained core.

The Australian context

Australian SaaS companies benchmarking against US metrics should note that Australian enterprise sales cycles are longer, which affects early cohort retention (users may take longer to fully onboard). Subscription e-commerce in Australia has lower baseline retention expectations than comparable US markets due to lower subscription culture penetration.

Where people get this wrong

Using aggregate retention metrics instead of cohort retentionAggregate retention (total active users / total users ever) is contaminated by growth rate. Fast-growing businesses with declining retention look fine on aggregate metrics until growth slows.

Making cohorts too granularDaily or weekly cohorts in small businesses produce noisy data. Monthly cohorts give more statistically meaningful signals.

Not annotating when product or experience changes happenedWithout annotations, you can see that retention improved in a cohort but not understand why. Track product changes, email sequence changes and channel shifts alongside cohort data.

Common questions

How do I run a cohort analysis in GA4?

In GA4, go to Explore > Cohort exploration. Set the cohort inclusion event (first visit, first purchase), the return criterion (any activity, specific event), and the cohort size (weekly or monthly). GA4 shows cohort retention tables and graphs.

What is a good cohort retention rate for e-commerce?

It depends heavily on product category and purchase frequency. Consumables (coffee, skincare) should see 30-40% 90-day repurchase rates. Considered purchases (furniture, electronics) have naturally lower rates. Compare to your own historical cohorts rather than abstract benchmarks.

What is the difference between cohort analysis and segmentation?

Segmentation groups users by characteristics at a point in time (age, location, product purchased). Cohort analysis tracks those groups over time to observe behaviour change. Cohort analysis is segmentation with a time dimension.

How long should I track cohorts?

At least as long as your average customer lifecycle. For monthly subscription products, track for 12 months. For considered purchases, track for the expected repurchase window (6-24 months). The goal is to see where the retention curve stabilises.

Keep exploring

About New Rebellion

New Rebellion is a marketing intelligence consultancy. We build tools, score Australian businesses on how their marketing actually performs, and publish Debrief every day. This dictionary is part of how we work in the open.

How we think →