Cohort Analysis Guide
Analyze customer cohorts to understand retention trends and optimize for lifetime value. Learn to spot patterns, predict churn, and identify your best acquisition channels.
What is Cohort Analysis?
Cohort analysis groups customers by a shared characteristic (signup date, acquisition channel, plan type) and tracks their behavior over time. This reveals patterns that overall averages hide.
- Spot retention trends: Are new cohorts stickier than old ones?
- Predict LTV accurately: Model future revenue by cohort behavior
- Identify best channels: Which sources produce highest-LTV customers?
- Measure product changes: Did that feature launch improve retention?
4 Types of Cohort Analysis (With Use Cases)
1. Time-Based Cohorts (Most Common)
Group customers by the month/week they first signed up or made their first purchase
Use Cases:
- • Track retention rate improvements over time
- • Measure impact of product launches or major changes
- • Identify seasonal patterns in customer lifetime value
- • Example: "January 2024 cohort has 45% 6-month retention vs 32% for January 2023"
2. Acquisition Channel Cohorts
Group by how customers found you (Google Ads, organic, referral, etc.)
Use Cases:
- • Calculate true CAC payback period by channel
- • Identify which channels drive highest-LTV customers
- • Optimize marketing budget allocation
- • Example: "Organic search customers have 3x higher LTV than paid social"
3. Behavioral Cohorts
Group by specific actions taken (completed onboarding, used key feature, etc.)
Use Cases:
- • Prove which activation events drive retention
- • Identify "aha moments" that predict long-term engagement
- • Guide product development priorities
- • Example: "Users who invite a teammate in first week have 85% retention vs 28%"
4. Segment/Attribute Cohorts
Group by customer characteristics (company size, industry, plan tier, geography)
Use Cases:
- • Identify your ideal customer profile (ICP)
- • Customize onboarding by segment
- • Focus sales efforts on high-LTV segments
- • Example: "50-200 employee companies have 2.5x higher LTV than 1-10 employee"
How to Build a Retention Cohort Table
Example: Monthly Retention Cohort
| Cohort | Size | Month 0 | Month 1 | Month 2 | Month 3 | Month 6 |
|---|---|---|---|---|---|---|
| Jan 2024 | 500 | 100% | 68% | 52% | 45% | 38% |
| Feb 2024 | 620 | 100% | 72% | 58% | 51% | - |
| Mar 2024 | 750 | 100% | 75% | 62% | - | - |
Key Insight: Retention is improving over time. March cohort has 75% month-1 retention vs 68% for January. Something changed for the better.
How to Calculate Each Cell:
Month 0: Always 100% (everyone starts active)
Month 1: (Active users in month 1) / (Total cohort size) × 100
Month 2: (Active users in month 2) / (Total cohort size) × 100
Note: Some analysts calculate retention as % of *previous month* instead of *original cohort*. Be consistent with your chosen method.
5 Patterns to Look for in Cohort Data
1. Improving Retention Over Time
Good Sign: Newer cohorts retain better than older ones
This means product improvements, onboarding changes, or better targeting are working. Double down on what changed.
2. Declining Retention Over Time
Warning Sign: Newer cohorts churn faster than older ones
Product-market fit may be weakening, or you're attracting lower-quality customers. Investigate immediately.
3. The "Smile" Curve
Pattern: Sharp drop in month 1-2, then flattening retention curve
Common in SaaS. Focus on getting users past the initial drop—those who survive 3 months often stay for years.
4. Channel Performance Gaps
Insight: 3x difference in retention between best and worst channel
Example: Organic search = 60% month-6 retention, paid social = 20%. Shift budget accordingly.
5. Seasonal Patterns
Pattern: Q4 cohorts consistently underperform Q1-Q3
Could indicate holiday signups are less serious, or year-end budget flush brings bad-fit customers.
Tools for Cohort Analysis
Analytics Platforms (Built-In Cohort Reports)
- • Amplitude: Best for product analytics, powerful behavioral cohorts
- • Mixpanel: Easy to set up, good for SaaS retention analysis
- • Google Analytics 4: Free, basic cohort reports (Engagement → Cohort exploration)
- • Heap: Auto-captures events, retroactive cohort creation
SQL/Data Warehouse (Custom Analysis)
- • BigQuery, Snowflake, Redshift: Full control over cohort definitions
- • dbt: Build reusable cohort models
- • Mode, Looker, Tableau: Visualize custom SQL cohort queries
Spreadsheets (For Small Datasets)
- • Google Sheets: Use COUNTIFS to build cohort tables
- • Excel: Pivot tables for quick cohort summaries
- • Good for <1,000 customers or manual analysis
5 Cohort Analysis Mistakes to Avoid
Need 100+ users per cohort for reliable insights. Combine weeks into months if needed.
Don't compare time-based cohorts to channel-based cohorts. Keep analysis consistent.
Be specific: "Logged in" vs "Performed key action" makes huge difference.
Users might stay but downgrade. Track MRR cohorts alongside retention.
Cohort analysis only matters if you use it to improve product, marketing, or onboarding.
Quick Start: Your First Cohort Analysis
Start with time-based (monthly signups). Easiest to implement.
SaaS: "Logged in and used core feature" | E-comm: "Made a purchase"
Need at least 6 months of data to see meaningful retention patterns.
Use GA4, Mixpanel, or a spreadsheet. Track month 1, 3, 6, 12 retention.
Improving retention? Double down. Declining? Fix onboarding or targeting.
Need Help Setting Up Cohort Analysis?
Our team will set up cohort tracking, build custom dashboards, and help you identify actionable insights to improve retention and LTV.
Get Your Free Funnel Audit