How practitioners model competition influences the predicted effects of a merger. We consider three models that are commonly used to evaluate horizontal mergers: Bertrand price setting, second score auction, and Nash bargaining. We first show how these models relate to one another, and specifically that the Bertrand and second score auction models can both be nested within a bargaining framework. Second, we show through numerical simulations how the predicted merger effects vary with model choice. Third, we show that two commonly used strategies for obtaining demand parameters can yield markedly different outcomes across the three models. Finally, we show how model and calibration strategy choices affect the magnitude of predicted harm in the 2012 Bazaarvoice/Power Reviews merger.