This article uses computational methods that reveal substantive differences among the equilibrium outcomes from three models of procurement competition: a recently developed model requiring numerical solution and two analytically tractable models that might naturally be considered suitable proxies. The models differ in what sellers know about the buyer's preferences for their products, and they yield substantially different prices and payoffs, different implications for institutional choice, surprising intensities of competition, and qualitatively different comparative statics. These findings caution against using the analytically convenient models when the newer model is empirically appropriate. Reinterpreting the models and results also provides new insights regarding price discrimination in oligopoly.