So, let's plug these shapes into a regressor / ranker / classifier, along with some possible target variables (career path, survival rates, yearly office peer ratings, salary after 5 years,) for as many balanced samples as possible, and see what predictive power these might have!

Or, I guess the graph itself could also be considered a resume differentiator / sales technique.

Edit: Oh, I see. They're heavy on the highly personalized matching angle.