Have you ever wanted to study how different subsets of your users behave over time? For example, ever wondered if users who register during a promo period have a higher average lifetime revenue than those who don’t? If the answer is ‘Yes!’, then the Cohort Report Builder is the perfect tool for you. Magento BI is specifically optimized to perform this analysis and make it relevant to your business.
Cohort analysis can be broadly defined as the analysis of user groups that share similar characteristics over their life cycles. It allows you to identify behavioral trends across different user groups.
For a more in-depth primer on cohort analysis, take a look here - we wrote the site on it!
In your Magento BI dashboard, it’s easy to create user cohorts based on a cohort date and a metric in your account.
Cohort analysis in action! Here, we can see revenue growing over time on a cumulative and per-user basis.
As mentioned above, cohort analysis allows you to identify behavioral trends among different user groups. With a solid understanding of how certain groups behave, you can tailor your decisions and spending to maximize your sales. Take, for example, a lifetime revenue cohort analysis - while this kind of analysis is beneficial for many reasons, the immediate one is better customer acquisition decisions.
In this section, we'll walk you through creating your own cohort analysis. For examples (and animated GIFs demonstrating the process), take a look at the Examples section of this article.
Now that we’re in the Cohort Report Builder, let's add the metric (ex: Revenue or Number of orders) that we want to perform the analysis on.
Note: Native Google Analytics metrics aren't compatible with the Cohort Report Builder. We recommend you reach out to support if you're trying to do this so we can help you brainstorm a solution.
The next step is to specify the cohort date. This is the date by which your users will be grouped. For example, this might be User’s first order date or User’s registration date.
Please note that you cannot use the same date the metric is built on (ex: created at) as the cohort date.
Next, we'll set the Interval and Time Period.
The Interval option allows you to set the length of your cohorts. For example, if this is set to Month, your report will be measured in months.
You can change how these intervals are displayed on the x-axis using the Duration menu.
Use the Time Period menu to choose the specific user cohorts to analyze. You can show every cohort, choose from a list, specify a time range, or define a rolling time range of cohorts to include.
For example, if we used the Specific Cohorts option, we can select specific months to include in the analysis:
If we were grouping our cohorts by registration date and then selected April, May, and June in the Specific Cohorts list, any users who registered in those months would be included.
Under duration, you can define the chart’s x-axis settings - that is, how many time periods each data point represents and how many data points to include in the analysis.
If you opted to group users by a cohort date that’s been joined from another table, you may see a counting members in the … table option:
Let's look at an example to understand this setting. Suppose you built a report cohorting a Revenue metric by Customer’s registration date. You also wanted to use the perspective Average value per cohort member to see the revenue per buyer over time. To find the average value per buyer, we need to decide on the number of buyers to divide by. Is it the number of registered customers in your customers table, or is it the number of distinct buyers in your orders table for the same period of time?
This setting answers that question. Counting members in the customers table includes all customers (whether or not they made a purchase, ever) in the average. Counting members in the orders table includes only customers who made a purchase.
After you’ve defined the metric and how you want to analyze it, you can select the perspective you want to use.
Just above the report visualization is a drop-down of perspective settings.
This shows the incremental contribution of a given cohort group at any given point in their life-cycle. (e.g.The “Week 6” point displays all data points made by users in their sixth week).
Average Value per Cohort Member
This divides the Standard cohort analysis in (1) by the number of users in each cohort group. This can be useful for comparing cohort performances on an apples-to-apples basis, as not all cohort groups may include the same number of users. For example, the average week 6 revenue per user from a certain cohort.
This perspective shows the traditional cohort analysis on a cumulative basis. In other words, it shows the total contribution of a given cohort to date at any given point in their life cycle. For example, the cumulative revenue after 6 weeks of users from a certain cohort.
Cumulative Average Value per Cohort Member
This divides the Cumulative analysis in (3) by the number of users in each cohort group. It shows the average lifetime contribution (often average lifetime revenue) per cohort member at each period in the cohort's life. For example, the average lifetime revenue after 6 months of users that joined in June.
Percent of First Value (show first value)
This analyzes the aggregate cohort contribution at a specific time in a cohort’s life cycle as a percentage of their contribution in the first period. For example, the month 6 revenue divided by the month 1 revenue of users that joined in June.
Percent of First Value (hide first value)
This is the same as the perspective above, except that the first time period value of 100% is hidden.
Now that we’ve gone through how to create a cohort analysis, let's take a look at some examples.
In this example, we analyzed the Revenue metric, grouped our cohorts by the customer’s first order date, and selected the 8 most recent cohorts (defined in the Time Period menu) to include in the analysis. To see how the cohorts grew over time, we used the Cumulative Average Value per Cohort Member perspective.
For this example, we analyzed the Number of orders metric, grouped our cohorts by the customer’s first order date, and included the 8 most recent cohorts (defined in the Time Period menu) in the analysis. To see the average number of orders for each cohort, we changed the perspective to Average Value per Cohort Member.
For our last example, we analyzed the Revenue metric, grouped our cohorts by the customer’s first order date, and included the 8 most recent cohorts (defined in the Time Period menu) in the analysis. To compare future purchasing activity to their first month, we set the perspective to Percent of First Value (hide first value).
The Cohort Builder is currently optimized for grouping users by a common cohort date. You might be interested in grouping the users by a similar activity or attribute - if that’s the case, we would love to help! We recommend checking out this tutorial on qualitative cohorts to get started.