Modeling the price of multi-attribute products generally requires an assessment of each attributes' market value. In the presence of price dispersion, when similar products are sold at different prices, hedonic pricing models provide users with biased estimates of attribute value. This paper develops the hedonic pricing literature by proposing data envelopment analysis as a prior means of identifying a sub-sample of products which, after adjusting for attribute provision, display no price dispersion. These products then display a homogenous link between attributes and price, which can be modeled using hedonic pricing. This paper implements and evaluates this two-stage approach using 1000 observations from the UK mortgage market.