Doing a shitty statistic unit of my own, your analysis all seems to be very good and checked out. The only thing i could say is instead of doing a t-test for each combination of the archetypes, you could look to do a between-group ANOVA since there is 3 or more groups where it will do it all at once, something minor but just wanted to let you know (less chance of error).
Other than than, really interesting and detailed article, these types of things are always awesome to look at.
Other than than, really interesting and detailed article, these types of things are always awesome to look at.