Marketing Analysis

Customer Segmentation through Cohort Analysis

The project strategically segments online sales customers using cohort analysis, providing insights into retention, purchasing patterns, and engagement

  • Date: 12 Feb 2023
  • Tools: Excel, Python
  • Area:
    marketing

Project Goals

The data is derived from real-life scenarios of an online company.

  • Assess Business Health: Increasing revenue, even without acquiring new customers, is a key indicator of a healthy business.
  • Understand Customer Behavior: Gain insights into customer behavior to inform strategic decision-making.
  • Boost Customer Retention: Create targeted marketing campaigns and personalized experiences through the effective segmentation of user groups and cohorts.
  • Boost Customer Retention: Analyze retention rates and identify potential churn risks to proactively enhance customer experiences.
  • Optimize App for Increased Interest: Identify trends and patterns in the customer lifecycle to optimize the user experience and increase customer lifetime value.

Cohort Analysis

Cohort analysis is indeed a descriptive analytics tool used to group customers into mutually exclusive cohorts, allowing for the measurement and comparison of metrics over time. the tool is valuable for comparing metrics across both the product lifecycle and customer lifecycles.

Types of Cohorts

Cohorts can be defined based on various criteria, and the type of cohort often depends on the specific context of analysis.

  1. Time-based Cohorts:´Grouping customers based on the time they first engaged with a product or service (e.g., monthly sign-ups).
  2. Behavior-based Cohorts: Grouping customers based on specific behaviors or actions they take (e.g., users who made a purchase within a certain timeframe).
  3. Size-based cohort: The diverse customer segments are categorized by their spending levels on the company's products or services within a specific timeframe after acquisition. This segmentation can be based on the total amount spent or the product type that constitutes the majority of their order amount during a defined period.

The cohort table is similar to a pivot table. It comprises three elements: the assigned cohort in rows, cohort index in columns, and metrics within the table.

How to perform a cohort analysis: steps and process

  1. Specifiy your goals

    2. The first step is to have clarity in your goals. What do you want to understand from the data that you have? What do you want to learn from this analysis and how will it help your business?

    Do you want to understand more about customer behavior, like how they landed on your website or how much time they spent on an app? Or do you want to understand at which point they are churning? These answers will help you identify which cohorts to track and which metrics to choose.

  2. Choose the metrices you want to measure

    3. Conversion rate

    Retention rate

    Average purchase value

    Customer lifetime value

    Engagement rate

    Churn rate

    Customer satisfaction

  3. Define relevant cohorts and gather data

    4. You need to consolidate relevant data like the date of first purchase or the total amount spent.

  4. Analyze and interpret data

    5. Once you have collected the data, you can use different methods of analysis like time series or regression to spot patterns. This will allow you to notice any critical changes that may have happened.


    Example: Time-based cohort analysis

    1. Segment customers into acquistion cohorts based on the month they made their first purchase.
    2. Assign cohort index to each purchase of the customer - the number of months since the first transaction (acqusition).
    3. Mark each transaction based on its relative time period since the first purchase.
    4. Calculate metrics, such as retetnion, average speed value.
    5. Build the heatmap

    • Look at the date
    python cohort

    Build cohort table

    • Create invoice months and cohort month

    The cohort month is defined as the acquisition month, which is the oldest invoice month of each customer.

    python cohort python cohort

    • Assign time offset value

    Creating a cohort index calculated in months and offset by 1 involves assigning each customer to a specific cohort based on the month they were acquired and then calculating the number of months since their acquisition, with an offset of 1.

    python cohort

    • Count monthly active customers from each cohort

    For each month after the acquisition, count the number of active customers in each cohort. An active customer is one who made at least one purchase in a given month.

    python cohort

    Cohort metrics

    • Customer retention rate

    Customer retention rate = (No. of customer at the end of a period - No. of customers acquired during that period )/ No. of customers at the start of the period * 100

    python cohort

    • Cohort quantity metric
    python cohort

    Visualizing cohort analysis

    python cohort python cohort

    Size-based cohort analysis: Daily turnover

    python cohort python cohort python cohort python cohort python cohort

    More Application of Cohort Analysis

    1. Customer Retention Analysis:

      Cohort analysis helps businesses track and understand customer retention over time. By grouping customers based on their acquisition or conversion dates, companies can assess how well they retain customers and identify patterns or trends in customer behavior.

    2. Product Performance Analysis:

      Businesses can use cohort analysis to evaluate the performance of specific products or services. By grouping customers who purchased a particular product within the same time frame, companies can assess the product's lifecycle, customer loyalty, and overall impact on revenue.

    3. Marketing Campaign Effectiveness:

      Cohort analysis allows marketers to assess the success of different marketing campaigns by grouping customers based on the time they were acquired through those campaigns. This helps in understanding which campaigns contribute most effectively to customer acquisition and long-term value.

    4. User Engagement and Churn:

      In the context of online platforms, cohort analysis is useful for tracking user engagement and identifying potential churn patterns. By analyzing cohorts of users based on sign-up dates, businesses can optimize user experiences and implement targeted strategies to reduce churn.

    5. Subscription Services Analysis:

      Cohort analysis is particularly relevant for subscription-based businesses. It helps in evaluating subscriber retention, understanding user behavior over subscription periods, and identifying factors influencing subscription renewals or cancellations.

    6. Employee Performance and Training:

      Beyond customer-related applications, cohort analysis can be applied to employee performance. For example, grouping employees based on their start dates to assess training effectiveness, retention, and performance over time.

    7. Financial Analysis:

      In finance, cohort analysis can be applied to evaluate the performance of financial products, investment portfolios, or the effectiveness of financial strategies. Cohorts may be based on the time of investment or the introduction of a financial product.

    8. E-learning and Education:

      Cohort analysis is beneficial in educational settings to analyze student performance, engagement, and course effectiveness. Cohorts can be formed based on enrollment dates or the introduction of new curriculum elements.

    9. Mobile App User Analysis:

      For mobile applications, cohort analysis helps in understanding user behavior, feature adoption, and retention. Cohorts based on app download dates can reveal insights into how different user groups engage with the app over time.

    10. Healthcare and Patient Outcomes:

      In healthcare, cohort analysis can be applied to study patient outcomes based on the time of diagnosis, treatment, or intervention. This can assist in improving patient care and treatment strategies.

    Next