One of the biggest problems that eCommerce companies face is inventory management. Having too little or too much inventory for a product lowers profitability. Lack of inventory can also mean a lost customer.
There are many solutions for inventory management today. Most of these focus on forecasting sales for existing products or newer versions of existing products. There are fewer and less proven methods for forecasting sales of completely new products. And understandably so - without any historical sales data for comparable products, it is more intuition than science.
You get your hands on a new data set. The goal - to get a deeper understanding of a specific phenomenon, say churn. You do the due diligence - eyeball the data to see what data points it contains and prepare some basic summaries (averages, medians etc.). Then you start asking questions of the data to get a better understanding of churn.
However, after building out a few reports, nothing stands out. The behavior of users who end up churning seems very similar to users who renew. Their interactions with your product/service are not very different. Customer support metrics are equally positive.
This leaves you confused and a bit apprehensive. If there are no visible differences or indicators, how do you optimize renewal?