This paper introduces algorithmic micro-foundations for formulating and solving strategic problems. It shows how the languages and disciplines of theoretical computer science, ‘artificial intelligence’ and computational complexity theory can be used to devise a set of heuristics, blueprints and procedures that can help strategic managers formulate problems, evaluate their difficulty, define ‘good enough solutions’ and optimize the ways in which they will solve them in advance of attempting to solve them. The paper introduces both a framework for the analysis of strategic problems in computational terms, and a set of prescriptions for strategic problem formulation and problem solving relative to which deviations and counter-productive moves can be specified and measured.