Cohort Analysis

Quick Start

Analyze how groups of users/customers (cohorts) behave over time, typically measuring retention, revenue, or engagement patterns.

Context Requirements

1. Dataset: User/customer event data
2. Cohort Definition: How to group users (by signup month, acquisition channel, etc.)
3. Retention Metric: What counts as "retained" (login, purchase, usage, etc.)
4. Time Periods: Analysis granularity (daily, weekly, monthly)

Context Gathering

Initial Questions:

"Let's set up cohort analysis. I need:

1. What are we analyzing?

   • User retention (returning users)
   • Revenue retention (recurring purchases)
   • Feature adoption (using specific features)
   • Other behavior

2. How should we define cohorts?

   • By signup date (most common)
   • By acquisition channel
   • By first purchase date
   • By product/plan tier
   • Other dimension

3. What counts as 'active' or 'retained'? Examples:

   • Logged in at least once
   • Made a purchase
   • Used feature X
   • Spent >10 minutes

4. What time periods?

   • Daily cohorts (for apps with daily usage)
   • Weekly cohorts
   • Monthly cohorts (most common for SaaS)
   • Quarterly cohorts"

For Dataset:

"I need data with:

• User ID (to track individuals)
• Cohort date (e.g., signup_date)
• Activity dates (e.g., login_date, purchase_date)
• Cohort attributes (optional: channel, plan, etc.)

Can you provide:

• File upload (CSV/Excel), OR
• Database query to fetch this, OR
• Description of tables and I'll write the query?"

Validation Questions:

"Before I proceed:

• What minimum cohort size should we analyze? (I recommend >100 users)
• How many periods should we track? (e.g., 12 months, 8 weeks)
• Any cohorts to exclude? (e.g., test users, employees)"

Workflow

1. Data Preparation