In product optimization, preference mapping techniques are widely used. Although their advantages and benefits are evident, these methods also show some limitations. For example, in external preference mapping (PrefMap), in general only two dimensions are used in the individual regression models, and there is no evidence that these dimensions are relevant for liking. Moreover, potentially important information which could explain the liking scores and which could be present on the third and higher dimensions of the sensory space is often not considered.For that reason, a new methodology, called PrefMFA, is proposed. This methodology based on Multiple Factor Analysis (MFA) is in between internal and external preference mapping since it takes the dimensions from the “common” product space between external (usually sensory) and the hedonic scores in the individual regressions. An extension to Hierarchical Multiple Factor Analysis (HMFA), when more than one external matrix is available, is also proposed.The PrefMFA methodology is applied to two datasets according to two different strategies of analyses, and the advantages of PrefMFA over PrefMap are shown. More precisely, the help for the interpretation as well as the various criteria proposed by MFA (such as the partial axes representation, the Lg, and the RV coefficients) help to better understand the strength of the underlying relationship between the external information and liking.