This article investigates the categorization of an ambiguous product (i.e., one that may be categorized in multiple ways) by applying insights from recent psychological and artificial‐intelligence research on concept representation. In particular, the article introduces a type of prototype representation (attribute‐value structure) that utilizes the notion of attribute diagnosticity. Attribute diagnosticity is defined as an attribute's usefulness for distinguishing instances from noninstances of a category and proposed as a determinant of an ambiguous product's typicality as a member of a category. A new measure of attribute diagnosticity is also developed based on a linear combination of cue validity and category validity. Attribute diagnosticity is shown to influence the perceived typicality of an ambiguous product. Several suggestions are provided for future use of attribute diagnosticity in research on product categorization.