Methods are provided for selecting a question to be presented to a user of an electronic product catalog system (1) to assist identification of a suitable product from a set of potentially suitable products according to the user's needs. The question is selected from a group of questions stored in the system (1). The methods use product data, which defines features of products in the product set and product scores associated with respective products in the product set, and rule data which defines rules relating answers associated with the questions to product feature constraints. The product data and rule data are processed to calculate question scores (Q_S) for respective questions such that the question score (Q_S) for each question is dependent on at least one of (a) the product scores of any products excluded from the product set if a rule relating to an answer associated with that question is effective, and (b) the product scores of any products retained in the product set if a rule relating to an answer associated with that question is effective. The question to be presented to the user is then selected in dependence on the question scores (Q_S). The disclosed methods allow questions to be selected to facilitate identification of a suitable product from an arbitrary initial product set, allowing needs-based interviewing to be conducted after feature-based filtering in catalog systems. Corresponding apparatus is also provided.