A method and a system for speeding up and reducing the latency in retrieving data of a data network (1). The network (1) being of the type from which data can be retrieved as data pages, among others, multi-level data pages from a plurality of data servers (2) and by a plurality of users (6; 16). Users are provided with an editable navigation tool (19) for retrieving data pages from the network (1). Data pages and data page levels retrieved by a user (6; 16) with the aid of the navigation tool (19) are analyzed and stored in a personal proxy (9; 11) associated with the user (6; 16) and featured by a cache memory (10; 12). The cache memory (10; 12) is refreshed by retrieving relevant data pages and data page levels from the data network (1). Data pages and data page levels requested with the aid of the navigation tool (19) are at first instance retrieved from the personal proxy (9, 11) and, if not available, from the data network (1).