As someone who works in data analysis, I often find myself having to explain how to read a cohort chart. For those unfamiliar, a cohort chart is simply a graphical representation of data that shows how different groups change over time.
While they may look daunting at first glance, understanding how to read a cohort chart can be incredibly helpful in making business decisions.
In this post, I’ll walk you through everything you need to know about reading and understanding cohort charts. By the end of it, you’ll have all the tools you need to make better-informed decisions for your business!
How to Read a Cohort Chart
A cohort chart is a graphical representation of data that is used to track a specific group of individuals over time. The chart is composed of two axes; the x-axis represents time, and the y-axis represents the number of individuals in the cohort.
What are Cohorts?
A group of people who share a characteristic is called a “cohort.” This term can refer to any characteristic of a group of people but is most often used to refer to a grouping of people based on when they signed up for a service.
When referring to non-time-dependent grouping, the word “segment” is used instead.
What is a Cohort Analysis?
A cohort analysis is a method of tracking and analyzing how groups of users perform over a given period of time.
If you want to compare how profitable your recent acquisitions are compared to those you gained last year, you can divide your user base into groups by the month of their first acquisition. You can then use this type of comparison to analyze your revenue growth.
This type of cohort allows you to see how users who were acquired in 2021 perform in their first month compared to how those who you acquired from 2020 did in theirs.
This is different than the revenue generated by looking at total sales every month.
For example, consider a hypothetical sales graph like this:
The graph shows the monthly revenue growth, which is trending upwards.
The blue bar in the chart represents the monthly revenue from users who signed up in that month. This does not display any positive trends, which could indicate an issue with your user acquisition strategy.
In 2017, we earned $1,000 in total, with $800 coming from new users who signed up. In 2018, we generated $1,650, but only $300 of that came from 2018’s January cohort.
The revenue generated from users over 50 years old is hiding the fact that younger users are making less money.
Cohort analysis is a very powerful way to understand the long-term health and growth of your business. This tutorial will explain how to analyze your customer data using a tool called a “cohort”.
What Can the Cohort Chart tell us?
A customer’s loyalty can be tracked using a cohort chart. This can help you identify potential issues in their life cycle.
Customer loyalty is one of the most important, but often overlooked, aspects of a business.
How Should I Read a Cohort Chart?
When looking at a cohort analysis, you should look at each column individually. This can help you identify potential issues in the users’ life cycle.
When looking at your cohorts, pay attention to the point where the shade of the same color changes. The bigger the difference between the two shades, the more significant the difference was between two points in time. This can help pinpoint problems in your user’s life cycle.
In an ideal world, all your clients from Month 0 would still be using your product in Month 12. Unfortunately, this is rarely the case.
If you look at the graph, you’ll notice that 3, 5, and 9 have issues.
If you’re not sure what you’re looking for, look at the columns with similar shades and lighten towards the next month.
This chart is a record of the number of times people have logged in over time. A quick glance suggests that the number of people logging in is increasing over time.
Looking at the chart, it appears that most visitors lose interest in the website after 3 months. There is a clear drop in retention after 9 months of use. The oldest customers have the most consistent usage.
When you’re viewing a cohorts graph, you can ignore the last month of each segment. This is because it represents the next month.
To become a loyal, long-term, and repeat buyer, a customer must make two purchases of the same product from the same store. The length of this time period is the difference between the date of their first and last purchase of the same item.
The first date represents the first purchase and the last date the most recent.
Take a look at the graph below and see if you can figure out anything.
All new customers are entered in the first column which represents the first month they purchased.
Of the 69 customers in 2010, 39% (27) were still active in 2011, 21% (15) in 2012, and so on.
Of the 69 customers that made the first purchase in 2010, no one purchased in 2011.
So, 19% of your customers will make their second purchase of your product 2 months from when they made the first. Of the customers who bought before 2011, 10% will make a purchase 5-6 months after the first order.
The chart also shows that there’s a sudden drop in sales for customers who first bought in June 2011. This could suggest that there is seasonality or a change in marketing strategy that is affecting them.
If you have information on the demographics of your client base, you may be able to understand any inconsistencies in the numbers. For instance, if you made a change in your marketing campaign in June 2011, this could explain the differences.
If your business is seasonal, then you might want to consider a 24-month graph instead of a 12.
It’s important to keep track of what you were doing at specific times. For instance, you might have started a PPC campaign.
You’re pleased because you’ve noticed a jump in your sales but are they from repeat customers or one-offs?
Cohort chart analysis can help you identify patterns over time, such as an increase in Month 0 that is reflected in Month 1.
Your customers may make another order about 2 months from when they first bought from you. This could be due to your product having a consumable component or it could be that you sell small accessories with your main product.
Or, it could just be that your product is faulty. For example, if you sell door hinges, maybe your oil dries up after a few months and your customers need to buy more. Adding some contextual information (e.g. “We sold 50,000 liters of hinge lubricant last year”) to the numbers often reveals interesting insights.
How to Perform Your Own Cohort Analysis
Most marketers use a tool like Stitch to combine their customer data for cohorts.
Step 1: Pull Raw Data
To conduct a cohort analysis, you need to export data from your database into a spreadsheet.
When analyzing customers’ purchasing habits, you want a table that lists each customer’s purchases.
Each customer record contains a unique ID, the date of the purchase, the amount, and when the customer first purchased.
The more data you have about your customers, the better. Include information such as their geographic location, first purchase, and any referrals they came from. This would give you a much better understanding of your customer base and allow you to segment your cohorts more effectively.
Each of these extra fields requires an additional attribute to your database. Luckily tools like Stitch can make this easy by allowing you to access all this information in the same table.
Step 2: Create Cohort Identifiers
Now that you’ve imported your customer data, it’s time to analyze it. To perform cohort analysis, you’ll want to compare the group of customers who purchased for the first time within a certain time frame.
We need to group our cohorts by when they purchased the product, so we’ll need to translate each “Date” into a “Year-Month”.
Step 3: Calculate Lifecycle Stages
Once we know which customers belong to which cohorts, we need to classify the lifecycle stage of each member.
Let’s say that on January 10 someone made the first purchase and on March 15 they made a second purchase. Their first transaction would place them in the “Month 1” lifecycle stage and their second purchase would be in “Month 3”.
When you have completed this task, your Excel table should look like the example shown
Step 4: Create a Pivot Table and Graph
A pivot table allows you to aggregate data across dimensions such as time, location, or product.
We want to create a single pivot table that summarizes transaction totals by time period and displays the results in one column per customer.
Its data can be displayed on a basic spreadsheet.
Now that you’ve learned the basics of cohort analysis, it’s time to get creative! Use these tips to start building your unique analyses.
Bonus Step: Data Perspectives
The graph above is a type of data analysis called a “cohort”. But it’s difficult to interpret in this fashion. Another way of viewing this data is as a “cumulative” value of each cohorts’ spend.
This graph allows you to see how each of your cohorts’ average spending changes over their lifetime. This can help you determine how much each of your different customers is worth to you.
To further help, you should then divide the data points for each group by the total number of members of that segment. This will give you a better idea of how each cohort performs relative to its size.
You can calculate your cohorts’ average values without the influence of their sizes by creating a second pivot table and dividing the calculated size of the group by the other.
If you’ve made it to the end of this post, congratulations! You now know everything there is to know about how to read a cohort chart. I hope that armed with this knowledge, you’ll be able to make better-informed decisions for your business. Thanks for reading!