In November 2015 I was asked to host a data blending breakout session for a Seattle Tableau Usergroup (SeaTUG) meeting. Data blending in Tableau is usually something I avoid at all costs. It is slower than joining in Tableau and I usually can avoid the need to blend through custom SQL. However, I had just built a customer churn model in Tableau that I decided to use data blending in. The data came from different data sources (Salesforce and our data warehouse on Amazon Redshift) and I needed to relate two dates on the same data axis.
I decided to bring in a date dimension (a table with a row for each date – 2010 through 2020 in this case – and information about each of those dates) and duplicate my salesforce data so I could blend one copy of the datasource on customer start date and the other copy on customer expiration. That technique allowed me to count customers that joined or churned in a given month in one view. This video blog shows a step by step approach to using this technique to create a simple customer churn model. Even though it is simple, this is what I used at my company as our first stab at tracking customer churn. For a more in-depth look at how you can build on this to get a more targeted and elegant churn model, view this fantastic post by Stephen H. Noble.