Customer Cohort Analysis
Model Scenario
Do you want to understand how customer behavior changes over time, identify key trends, or measure the long-term impact of your business decisions? The Customer Cohort Analysis model in Graphite Note helps you do exactly that by grouping customers based on their first interaction or shared characteristics, and tracking their behavior over time.
A cohort is a group of customers who share a common attribute, most commonly the time of their first purchase. This model enables you to analyze how these cohorts perform over time, such as how often they return, how much they spend, and when their engagement drops off. This is a powerful tool for identifying growth opportunities, evaluating retention strategies, and improving customer lifetime value.
Setting Up the Model
Creating a Customer Cohort Analysis in Graphite Note is straightforward. Here’s what you need to configure:
Time/Date Column: Select a time-based column (e.g., order date or signup date) to define when customers enter a cohort.
Aggregation Level: Choose how to group the data over time—monthly, weekly, or daily. For example, selecting monthly will track cohorts by the month they made their first purchase.
Customer ID & Transaction ID: These are required. Customer ID identifies unique customers, while Transaction ID (Order ID) helps track purchases.
Monetary Column: Select a column that represents the monetary value (e.g., total amount spent) for each transaction. This is the key metric used to evaluate customer behavior.
Optional Breakdown (RepeatBy): You can break down cohort analysis by a business dimension (e.g., region, product category) by enabling the Repeat by option. This is available for variables with fewer than 20 unique values.
Once you’ve configured the settings, your Customer Cohort model is ready to run.
Model Results
The results of your cohort analysis are divided into three tabs: Cohorts, Repeat by, and Details.
Cohorts
This tab displays heatmaps that show how different customer cohorts behave over time. You can switch between several metrics.
Results representation
For each metric in the Cohort Analysis (like Amount, Percentage, or Number of Customers), the results are shown in two formats to help you better understand your customer behavior over time:
Line Chart (on top): Shows how each cohort (group of customers that started in the same period) performs over time. Each line represents one cohort, and the chart helps you quickly spot trends — for example, how spending or retention drops off or grows.
Cohort Table with Heatmap (below): Shows the exact values for each cohort and time period (e.g., Year 0, Year 1, Year 2…). The color intensity makes it easy to see where values are high or low — helping you spot strong-performing or weak-performing cohorts at a glance.
Together, these two visuals give a complete picture: a bird’s-eye trend view and detailed numbers in one place.
Metrics
Percentage: This metric shows you, in percentages, how many people from each group (cohort) are still active in the months or years after their first purchase — compared to how many started in that group. For example, let’s say 1,000 people made their first purchase in January 2022. If 300 of them came back and purchased again in February, the percentage shown for that month would be 30%. If only 150 came back in March, it would show 15%.
Number of Customers: Shows how many customers from each cohort made repeat purchases in subsequent time periods. Imagine a group of people who all made their first purchase in 2018. The chart then shows how many of them continued to buy again in the years that followed. For example, in 2019, a portion of that 2018 group returned and made another purchase. In 2020, a smaller part came back again, and so on. This pattern helps you understand how long people stay active after their first order and how engagement drops or changes over time for each starting group.
Amount: This metric shows how much money each group of customers (cohort) spent in each time period after their first purchase. For example, you can see how much the 2018 cohort spent in their first year, then how much they spent in the second year, and so on. It helps you understand how spending behavior changes over time, do customers keep buying, or does spending drop off?
Cumulative Amount: Instead of showing how much was spent in each individual year, this view adds it all up across the years. It tells you the total amount each cohort has spent from their first year up to the current period. This way, you can track the long-term value of each customer group and compare which cohorts brought in more total revenue over time.
Average Order Value / Revenue Per Customer: This metrics looks at how much revenue each customer brings on average. It’s especially useful when you want to compare the quality of cohorts, not just their size. For example, even if two groups have the same number of customers, one may bring in more revenue per person. This helps you identify high-value customer groups and refine your marketing or sales strategies.
Repeat By
If you enabled the Repeat by option when setting up the model, this tab will show separate cohort analyses for each value in the selected variable (e.g., each country or product type). This allows you to explore cohort behavior within specific segments of your business.
Details
All numerical results from the Cohorts and Repeat by tabs are available here in table format. You can export this data for further analysis, reporting, or dashboard integration.
Take actions with Customer Cohort Analysis
Uncover patterns
Customer Cohort Analysis helps you go beyond basic reporting by uncovering long-term behavioral patterns across customer groups. Once the model is trained, you can use it to answer important questions about customer retention, spending habits, and lifecycle performance.
Here’s how you can take action with Customer Cohort Analysis in Graphite Note:
Track Retention Over Time: Understand how many customers from each cohort return in subsequent periods. This helps evaluate how well your business retains users and where drop-offs occur.
Evaluate Monetization Strategies: See how much each cohort is spending over time. You can compare cohorts to understand whether recent strategy changes (like pricing, discounts, or onboarding) are improving lifetime value.
Spot Trends and Anomalies: Use the visual charts to identify changes in customer behavior. Are newer cohorts spending less or more than older ones? Are there periods of rapid drop-off? Use these insights to adapt your customer journey.
Compare Segments with “Repeat By”: Analyze how different customer segments (like product category, region, or channel) behave over time. This allows for fine-tuned marketing and retention strategies targeted at specific groups.
Optimize Remarketing Timing: By identifying when customer engagement typically drops off, you can time your outreach campaigns to re-engage users right when it matters most.
By analyzing how customer behavior evolves over time, Customer Cohort Analysis turns your raw transaction data into clear, actionable decisions that improve retention, increase revenue, and sharpen your business strategy.
Create Notebooks
You can share your prediction results with your team using the Notebook feature.
Notebooks allow you to create various visualizations with detailed descriptions. You can plot model results for better understanding and enable users to make their own predictions. For more information, refer to the Data Storytelling section.
Last updated
Was this helpful?