Learn Cohort Analysis | Segmentation and Behavioral Analysis

Swipe to show menu

Cohort analysis is a powerful technique in product analytics that allows you to compare groups of users who share a common starting point - such as their signup month or first purchase date. Imagine you run an app and want to understand how users who joined in January behave over time compared to those who joined in February. Rather than averaging all users together, cohort analysis lets you track each group's retention and engagement as they progress through their lifecycle.

Think of a cohort as a graduating class in school: all students who started in the same year experience their journey together, and you can observe how many remain at each milestone. In product analytics, this means you can see whether users from certain months stick around longer, engage more, or drop off at different rates.

For example, you might notice that users who signed up in February have higher week 4 retention than those from January. This could indicate successful product changes, seasonal effects, or differences in acquisition channels. By breaking down users into cohorts, you gain a clearer picture of how product updates, marketing campaigns, or external events impact specific groups over time.

Definition

A cohort is a group of users who share a common characteristic, such as signup month.


              1234567891011121314151617181920212223242526272829303132333435
            
import pandas as pd

# Example user data: user_id, signup_date, activity_date
data = [
    {"user_id": 1, "signup_date": "2024-01-10", "activity_date": "2024-01-10"},
    {"user_id": 1, "signup_date": "2024-01-10", "activity_date": "2024-01-17"},
    {"user_id": 2, "signup_date": "2024-01-15", "activity_date": "2024-01-15"},
    {"user_id": 2, "signup_date": "2024-01-15", "activity_date": "2024-01-22"},
    {"user_id": 3, "signup_date": "2024-02-05", "activity_date": "2024-02-05"},
    {"user_id": 3, "signup_date": "2024-02-05", "activity_date": "2024-02-12"},
    {"user_id": 4, "signup_date": "2024-02-20", "activity_date": "2024-02-20"},
    {"user_id": 4, "signup_date": "2024-02-20", "activity_date": "2024-02-27"},
]

df = pd.DataFrame(data)
df["signup_month"] = pd.to_datetime(df["signup_date"]).dt.to_period("M")
df["activity_week"] = (
    pd.to_datetime(df["activity_date"]) - pd.to_datetime(df["signup_date"])
).dt.days // 7

# Keeping only the first activity per user per week
df_cohort = df.drop_duplicates(subset=["user_id", "activity_week"])

# Counting users in each cohort and week
cohort_pivot = (
    df_cohort.groupby(["signup_month", "activity_week"])["user_id"]
    .nunique()
    .unstack(fill_value=0)
)

# Calculating cohort sizes (week 0)
cohort_sizes = cohort_pivot[0]
retention = cohort_pivot.divide(cohort_sizes, axis=0)

print(retention)

Interpreting cohort analysis results can reveal valuable insights for your product strategy. If you see that more recent cohorts have better retention, it could mean your latest features or onboarding improvements are working. Conversely, a sudden drop in retention for a specific cohort might highlight issues with a new release or a change in marketing tactics.

Cohort analysis helps you move beyond surface-level metrics and understand the real impact of product changes on user behavior. By tracking each cohort's journey, you can identify which strategies drive long-term engagement and retention, and where you may need to adjust your approach to keep users coming back.

Everything was clear?

Thanks for your feedback!

Section 3. Chapter 2

Ask AI

Ask anything or try one of the suggested questions to begin our chat

Section 3. Chapter 2