Informatics in Primary Care (2002) 10: 147–59 © 2002 PHCSG, British Computer Society Introduction There is a considerable literature on website metrics, but very little, if any, on digital interactive television (DiTV) metrics. This is not surprising given the relative novelty of DiTV, but given the rapid growth and importance of this medium, metrics will soon have to be developed and pressed into service. In fact, DiTV pages are stored and delivered to the user in much the same way as web pages, hence much of the research conducted on web metrics is applicable to the DiTV environment. Web metrics themselves have been developed over a relatively short period of time. 1–4 Indeed, most of the metrics are not wholly new and their origins can be traced back to research on OPAC (Online Public Access Catalogue) log files. 5,6 Typically, the metrics used are the number of pages or screens viewed (page impressions), number of users/ visitors, time spent viewing a page, number of pages viewed per search session, session time length, and the number of screens printed or downloaded. 4,7,8 Time measures are calculated from the date and time stamp field, and pages viewed by counting and verify- ing transaction lines. DiTV logs do follow the Internet conventions of recording the time that the client requests the page, although a log-off time is not recorded; as a con- sequence, this introduces problems when dealing with highly skewed data. There are, however, important Refereed papers Comparing two digital consumer health television services using transaction log analysis Paul Huntington BSc Research Fellow David Nicholas PhD MPhil ALA Head of Department of Information Science Peter Williams MSc BA Research Fellow Ciber, Department of Information Science, City University, London, UK Barrie Gunter Head of Research, Department of Journalism, University of Sheffield, Sheffield, UK ABSTRACT Use is an important characteristic in determining the success or otherwise of any digital information service, and in making comparisons between services. The source of most use data is the server logs that record user activity on a real-time and continuous basis. There is much demand from sponsors, channel owners and marketing depart- ments for this information. The authors evaluate the performance of use metrics, including reach, in order to make comparisons between two services and discuss the methodological problems associated with making such comparisons. The two services were: Living Health, managed by Flextech and distributed by Telewest, and NHS Direct Digital, managed by Communicopia Data and distributed by Kingston Interactive Television. The data were collected over the period August 2001 to February 2002. During this period, the two sites were visited by approximately 20 000 people who recorded more than three-quarters of a million page views. Keywords: digital television, health information, interactive channel, transaction log