Decision trees (DTs) are widely used to represent control strategies, e.g., for machine learning classifiers or formal verification results. Decisions in DTs are based on expressive predicates, but they seem to not fully exploit their potential towards concise representations. One reason is in their tree structure, leading to isomorphic subtrees not being merged. Reduced ordered binary decision diagrams (BDDs) inherently support merging, but allow only for reasoning on boolean variables with an ordering imposed. In this talk I will report on ongoing work comparing decision trees and decision diagrams w.r.t. their capabilities to concisely represent control strategies and how to overcome their drawbacks.