WEBSITE USER PROFILING USING ANONYMOUSLY COLLECTED DATA

First contextual data characterizing behavioral attributes of a user visiting at least one web page is received. The first contextual data is collected by anonymously tracking interaction of the user with the at least one web page via a data collector embedded in the at least one web page. Second contextual data characterizing non-behavioral attributes of the user is also received. The second contextual data is based solely on anonymously collected information and it originates from a data source other than the data collector. Thereafter, it is determined which attributes among a plurality of pre-defined attributes are present for the user based on the first contextual data and the second contextual data. The determined plurality of attributes are associated with a best fit amongst a plurality of clusters or associations generated by grouping users with similar attributes that participated in a plurality of historical transactions. Subsequently, using at least one predictive model trained with historical user conversion data, it is determined which of a plurality of available offers for presentation via at least one web page is most likely to result in a conversion of the user based on the pertinent cluster or association so that the presentation of the identified available offer can be initiated. Related apparatus, systems, techniques and articles are also described.