Mass customization requires consumers to construct their own version of the product via a series of attribute decisions (Dellaert and Stremersch, 2005; Franke, Keinz, and Steger, 2009; Franke, Schreier, and Kaiser 2010). To facilitate mass customization, options are organized such that similar attributes get put under the same category, i.e. they are organized in a taxonomic category format. However, recent research in cognitive psychology has shown that there may be other ways of categorizing the same information (Estes, 2003a; Golonka and Estes, 2009). For example, Blockbuster categorizes movies by genre (romance, action), release date (new releases, classics), or theme (holiday, sport); at Hertz traditional vehicle categories (sedan, truck) as well as thematic categories are used (fun, luxury, green, see also Ratneshwar, Barsalou, Pechmann, and Moore, 2001; Poynor and Wood, 2009).The objective of the present research proposal is to establish how different ways of categorizing the same information influence both the process of customization (satisfaction with the customization process and customization effort), as well as the outcome of the customization process (the novelty and usability of the customized product). Our theoretical lever for addressing this question is cognitive research on similarity. Traditionally, two things are similar if they possess the same features (Tversky 1977). However, the most recent cognitive research has pointed to an additional type of similarity, called thematic similarity (Estes et al. forthcoming; Estes, 2003a; Golonka and Estes, 2009). Two things are thematically similar, if they co-occur in a common scenario. A wealth of evidence consistently demonstrates the distinction between the two bases for categorizing information and points to differential effects that cohere on various levels of analysis (Estes, 2003a,b; Sass, Sachs, Krach, and Kircher, 2009). Despite this, little research in marketing and consumer behavior has investigated the effects of the two different bases for categorization. Even more problematically, the scanty results available to date remain inconclusive, and it is currently not clear which category format (taxonomic or thematic) might be more promising (e.g. Gibbert and Mazursky, 2009; Poynor and Wood, 2009; Noseworthy, Finlay and Islam, 2010). As such, we investigate when, i.e. under what conditions a thematic category format influences the relevant dependent variables (customization process as well as the customized product). Specifically, we draw on literature in cognitive science as well as our own prior empirical work to point a contingency model involving four moderating variables on the individual level and contextual level. Individual-level moderators include a person’s disposition towards a specific type of similarity when categorizing (thematic vs. taxonomic), as well as an individual’s experience with the category (experienced vs. novice). Contextual-level moderators include processing mode (holistic vs. piecemeal), and mood (positive vs. negative). Results from the proposed research would extend beyond recent work in mass customization, e.g. Levav et al.’s (2010) taxonomic model. The present research would further allow gathering insights how the attributes in a mass-customization system need to be assorted in order to increase consumers´ satisfaction with the configuration process itself and also how the preference for the customized product may be increased, two key outcome variables in mass customization (Franke, et al., 2010).