Building profiles of registered users of a web site, as in case of a portal, is of importance if it goes beyond collecting the obvious information the user is willing to give at the time of registration. The starting point of this work is the assumption that a community of users is registered on a web site and that for each user a profile is built. Such a profile contains two parts of data: one obvious, given directly by the user and one less obvious, inferred by the user?s behavior during his visits on the site. This paper presents ideas on how to build a user profile based on non-obvious information and takes into account various factors influencing in its development. Special emphases are paid on feedback mechanism and several methods of measuring its results.