As I turn my attention to potentially sitting the expert this was an interesting one. I need to know the predictive pallet a lot better but this question itself was not logically difficult at all. Just a case of knowing what tools can get you to the desired outcome.
Quick google search helped me realize I needed a forest model to get the chart to find the 10 variables to look at next.
In terms of the Chi-Square I wasted time doing another forest model before realizing that was not giving the value I wanted - a problematic time waster for the real deal. The nested test tool I had no knowledge about but is a good one to know to get the desired outcome in this case. Logistic regressions makes sense given we are predicting a non continuous target variable.