We analyse dynamic general equilibrium models with more-or-less directed search by informed buyers and random search by uninformed buyers. This nests existing specifications and generates new insights. A quantitative application concerns the welfare cost of inflation, which is known to be quite high with pure random search and low with pure directed search. Our calibration implies the impact of inflation is fairly low, in part because, in addition to the usual costs, it provides benefits by more heavily taxing high-price sellers that inefficiently profit from exploiting the uninformed. Other applications analyse analytically and numerically changes in credit conditions and information.