Learn Calculating Retention and Churn Metrics | Advanced Cohort Segmentation and Retention Metrics

Swipe to show menu

Retention and churn metrics are essential tools in cohort analysis, helping you measure how well your product retains users over time and where you may be losing them. The retention rate quantifies the percentage of users from a cohort who remain active after a given period. The churn rate is the complement, showing the percentage of users who have stopped engaging. The survival rate tracks the probability that a user remains active up to each time period, providing a view of user longevity.

Formulas:

Retention Rate (at period n):
Retention Rate = (Number of users active in period n) / (Number of users in cohort at period 0);
Churn Rate (at period n):
Churn Rate = 1 - Retention Rate (at period n);
Survival Rate (at period n):
Survival Rate = (Number of users still active at period n) / (Number of users in cohort at period 0).

These metrics are often visualized using retention curves or survival plots, which help you quickly spot patterns, such as steep drop-offs or periods of stability. By tracking these rates across multiple cohorts and time periods, you can identify successful engagement strategies and areas needing improvement.


              123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566
            
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# Example cohort data: each row is a user, columns are activity in each month (1 = active, 0 = inactive)
data = {
    "user_id": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
    "cohort_month": ["2023-01"] * 5 + ["2023-02"] * 5,
    "month_0": [1, 1, 1, 1, 1, 1, 1, 1, 1, 1],  # All users active at signup
    "month_1": [1, 0, 1, 1, 0, 1, 1, 0, 1, 1],
    "month_2": [1, 0, 0, 1, 0, 1, 0, 0, 1, 0],
    "month_3": [0, 0, 0, 1, 0, 1, 0, 0, 0, 0],
}

df = pd.DataFrame(data)

# Calculating retention, churn, and survival rates for each cohort
results = []
for cohort, group in df.groupby("cohort_month"):
    cohort_size = len(group)
    retention = []
    churn = []
    survival = []
    users_remaining = cohort_size

    for month in ["month_0", "month_1", "month_2", "month_3"]:
        active = group[month].sum()
        retention_rate = active / cohort_size
        retention.append(retention_rate)
        churn.append(1 - retention_rate)
        survival.append(users_remaining / cohort_size)
        users_remaining = active  # Update for next period

    results.append({
        "cohort_month": cohort,
        "retention": retention,
        "churn": churn,
        "survival": survival,
    })

# Converting results to DataFrame for plotting
metrics_df = pd.DataFrame(results)
months = ["month_0", "month_1", "month_2", "month_3"]

# Plot retention curves
plt.figure(figsize=(10, 5))
for idx, row in metrics_df.iterrows():
    plt.plot(months, row["retention"], marker="o", label=f"Cohort {row['cohort_month']}")
plt.title("Cohort Retention Curves")
plt.xlabel("Months Since Signup")
plt.ylabel("Retention Rate")
plt.legend()
plt.show()

# Plot survival curves
plt.figure(figsize=(10, 5))
for idx, row in metrics_df.iterrows():
    plt.plot(months, row["survival"], marker="o", label=f"Cohort {row['cohort_month']}")
plt.title("Cohort Survival Curves")
plt.xlabel("Months Since Signup")
plt.ylabel("Survival Rate")
plt.legend()
plt.show()

# Printing calculated metrics
print(metrics_df[["cohort_month", "retention", "churn", "survival"]])

This code demonstrates how to calculate and visualize retention, churn, and survival rates for user cohorts using pandas and matplotlib in Python.

The purpose is to analyze how groups of users (cohorts) behave over time, focusing on their engagement and longevity with a product.

Data Structure:

The data is organized as a DataFrame where each row represents a user;
Columns include a unique user ID, the cohort month (when the user joined), and binary activity indicators for each month (1 = active, 0 = inactive).

Calculation Logic:
For each cohort, the code:

Calculates the retention rate for each month as the proportion of users still active compared to the original cohort size;
Computes the churn rate as the complement of retention (1 - retention rate);
Tracks the survival rate, which shows the probability that a user remains active up to each period, updating the count of remaining users after each month.

Visualization:

Retention and survival curves are plotted for each cohort using matplotlib;
These plots help you visually compare how quickly users drop off (churn) or remain engaged (retention/survival) across cohorts and time periods, revealing trends and patterns in user engagement.

Interpreting retention and churn metrics is crucial for making informed business decisions. High retention rates indicate that users find ongoing value in your product, suggesting strong engagement and product-market fit. Conversely, high churn rates may signal issues such as unmet user needs, poor onboarding, or competitive pressure. Survival curves help you visualize how quickly cohorts are shrinking - steep drops may reveal when users typically lose interest.

By regularly tracking these metrics, you can identify trends over time and across different cohorts. For example, if a new feature rollout coincides with increased retention, it may be worth investing further in that direction. On the other hand, if churn spikes after a particular update, it could indicate a need for product improvement or additional user support. Ultimately, these metrics empower you to test hypotheses, optimize user journeys, and allocate resources effectively.

Everything was clear?

Thanks for your feedback!

Section 2. Chapter 2

Ask AI

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

Section 2. Chapter 2