Current theories argue that human decision making is largely based on quick, automatic, and intuitive processes that are occasionally supplemented by slow controlled deliberation. Researchers, therefore, predominantly studied the heuristics of the automatic system in everyday decision making. Our study examines the role of slow deliberation for experts who exhibit superior decision-making outcomes in tactical chess problems with clear best moves. Our study uses advanced computer software to measure the objective value of actions preferred at the start versus the conclusion of decision making. It finds that both experts and less skilled individuals benefit significantly from extra deliberation regardless of whether the problem is easy or difficult. Our findings have important implications for the role of training for increasing decision making accuracy in many domains of expertise. (Contains 3 figures.)