Cohort Analysis: What It Is and How to Use It

What Is Cohort Analysis

Cohort analysis is an analytical technique that groups users who share a common characteristic or experience within a specific time period, and then tracks their behavior over time.

A cohort can be as simple as "all users who signed up in January 2026" or as specific as "users who arrived from a Black Friday campaign and made their first purchase in the premium category."

The fundamental difference from aggregate metrics is that cohort analysis lets you see the temporal evolution of each group, revealing patterns that overall averages hide.

Why It's Essential for CRO

Detect Retention Problems Early

If the January cohort retains 40% of users at 90 days but the March cohort only retains 25%, something changed. It could be a shift in traffic source, an onboarding problem, or a product update that worsened the experience.

Measure the Real Impact of Changes

When you launch a checkout improvement, aggregate metrics mix old users (who already had formed habits) with new users. Cohort analysis lets you compare the "post-change" cohort with previous ones and isolate the real effect of the improvement.

Validate Traffic Quality

Not all traffic is equal. Cohorts by acquisition source show you which channels bring users who actually convert and stay, not just those generating the most volume.

Types of Cohorts

Acquisition Cohorts

Group users by the date they performed their first action (registration, first visit, first purchase). These are the most common and what GA4 uses by default.

Behavioral Cohorts

Group users by a specific action: completing onboarding, using a specific feature, reaching a spending threshold. More powerful for understanding engagement but require more advanced configuration.

Campaign Cohorts

Group users by the campaign or source that brought them. Essential for evaluating traffic quality from different channels and campaigns long-term, beyond the first conversion.

How to Run Cohort Analysis in GA4

Step 1: Access the Report

In GA4, go to Explore > Cohort Exploration. Select:

• Inclusion criteria: the action that defines the cohort (first visit, first purchase, etc.).
• Return criteria: the action that measures retention (any visit, purchase, specific event).
• Granularity: daily, weekly, or monthly.

Step 2: Configure Metrics

The most useful metrics for CRO are:

• Retention rate: percentage of users who return in each period.
• Transactions: number of purchases per cohort.
• Revenue: monetary value generated by each cohort over time.

Step 3: Segment

Apply segments to compare cohorts by:

• Device (desktop vs. mobile).
• Acquisition source.
• Country or market.
• First purchase product or category.

How to Read a Cohort Table

A typical cohort table looks like this:

Cohort	Week 0	Week 1	Week 2	Week 3	Week 4
Jan 1-7	100%	35%	22%	18%	15%
Jan 8-14	100%	38%	25%	20%	17%
Jan 15-21	100%	30%	18%	12%	10%

Key readings:

• Week 0 to Week 1 drop: measures "activation." If very steep (>70%), the onboarding or first experience has problems.
• Stabilization: the point where the curve flattens indicates your loyal user base. If it never stabilizes, you have a product problem.
• Comparison between cohorts: if one cohort has significantly different retention, investigate what changed during that period.

Derived Metrics from Cohort Analysis

N-Day Retention Rate

The percentage of users who return after N days. Benchmarks vary by industry:

• SaaS: 30-day retention of 40-60% is considered good.
• Ecommerce: 90-day retention of 20-30% is acceptable.
• Mobile apps: 7-day retention of 20% is already notable.

Revenue per Cohort

Total revenue generated by each cohort over time. Lets you calculate real LTV by acquisition channel and adjust your marketing investment.

Payback Period

The time it takes for a cohort to generate revenue equal to the acquisition cost. If payback lengthens in recent cohorts, it may indicate you're attracting lower-quality traffic.

Common Mistakes

Insufficient Cohort Size

Cohorts that are too small produce noisy results. As a general rule, you need at least 100-200 users per cohort for reliable data. If your traffic is low, use monthly instead of weekly granularity.

Ignoring Seasonality

A December cohort in a fashion ecommerce will behave very differently from a February one. Don't compare cohorts from different seasons without context.

Measuring Only the First Conversion

A cohort's value reveals itself over time. If you only look at the initial conversion, you miss the most important information: who comes back and who generates recurring value.

Practical Application in CRO

The workflow for integrating cohorts into your optimization process:

1. Establish a baseline: measure current retention by weekly cohorts for 2-3 months before making changes.
2. Identify the drops: locate at which point in the journey you lose the most users.
3. Optimize that point: if the drop is between the first and second visit, work on the welcome email, homepage personalization, or re-engagement triggers.
4. Measure impact by cohort: compare post-change cohorts with the baseline to validate that the improvement is real and sustained.

---

At Boost, we use cohort analysis as a fundamental part of our CRO process to understand not just whether users convert, but whether they stay. If you want to deepen your understanding of retention and customer value, learn about our CRO services or analyze your site for free with Scan&Boost.

Adrià Vidal — Boost