Leap day post! I'll celebrate my blog's birthday every 4 years! This is the first blog of my Cohort Analysis series, I can't wait to start the adventure with you!
Are you still struggling with how to retain your customers? Understand them first! If an organization is selling insurance products, they may want to know in which period, most customers who join in a plan were renewing it the next year. Analyzing purchase history can help the customer acquisition team to identify the key factors that make a group of people who have similar behaviors commit to the brand.
“A cohort is a group of subjects who share a defining characteristic”. “Cohort analysis is a subset of behavioral analytics that takes the data from a given data set and rather than looking at all users as one unit, it breaks them into related groups for analysis.” --Wikipedia
Visualization is a good way to reflect the purchase history of different cohorts.
In this blog, we will explore how to perform a cohort analysis in Tableau to see the share of customers acquired in a year were retained in the next couple of years.
🤪Yuchen’s tip: Always make full preparation before an analysis!
Plan the analysis:
1. Understand "Level of Detail (LOD) Expression" in Tableau
I’ll give a brief introduction to the “Fixed” function which is used for the cohort analysis today.
The picture above shows the level of “Fixed” is higher than the Dimension Filters, which means that a more detailed dimension in the view won’t affect the calculated result of “Fixed”. I created some calculation of different LOD in the example below, you’ll find the difference among them.
2. Explore the Dataset
The "Store sales" data set collected sales information from 2010 to 2013. The table shows some of the field names. It’s pretty straightforward and ready for analysis.
3. Clarify the Goal of Analysis
discovering whether customers who made the first purchase in a certain year (cohort),
continued to purchase in the next few years.
4. Identify the Type of Chart
You can picture the result in mind, the solution can be using the bar chart with colored cohorts and mark labels to show the difference of shares in each year.
Visualization is easy in this example, we'll focus more on the ideas of the analysis. Let's give it a try.
Step 1: Create a cohort
We define our cohort in this analysis as “year of acquisition”, so we can use the “Min” function to find the minimum purchase date or first purchase date for each unique customer.
Step 2: Create a Bar Chart
In the bar chart, we would see the number of customers who made purchases each year and break them down by the cohort in color.
Then You may add a tooltip to help people better interact with the dashboard:
Step 3: Exploring insights
The data of four years may not give you enough information to see more details, but we can still get some ideas from this simple example. 44.15% of customers acquired in 2010 made purchases in 2011 and only 38.8% of customers acquired in 2011 made purchased in 2012; 59.12% of a customer acquired in 2010 made purchases in 2013 which is the 3rd year after their first purchase. We may want to know if there’re new product features launched in 2010 that worked for most customers and they want to commit to the brand.
Acquiring new customers is expensive, companies want to retain loyal customers to bring them more profit. Exploring the historical data is the first step to help identify the special changes in a certain period that help to convince more customers committed to the product.
🤪Yuchen's Alert: This is not the end, I’ll see you in my next Cohort Analysis!
Happy analyzing!
留言