We propose a method of building control groups for retrospective merger evaluation. The procedure involves two parameters: a distance or radius, and an ‘order of exposure’ reflecting the notion that a merger can affect an outlet in a more or less direct way. We implement the method of estimating the price effect of a merger in the Paris parking market and illustrate the consequences of including indirectly exposed outlets in the control group. The methodology applies to any industry with spatial competition and allows one to check the sensitivity of the results to the choice of the control group.