The ability to create new predictive models, run them against historical data sets in order to understand how they perform over time (and then use that knowledge when creating future predictions). Also being able to see what is working well so we can continue improving it! I wish there was more documentation available online about this product as some of our team members are not familiar with Python or R programming languages which would make using these tools easier than if you had someone who knew both those things already. We have been solving problems related to predicting customer churn by running machine learning algorithms like decision trees/random forests etc., but also looking at other factors such as age groups within customers where certain products might be better suited depending upon their needs.