During Web Analytics Wednesday on 25 March 2009, the results of the study of differences in website statistics were presented. The conclusions have resulted in practical guidelines for determining the extent to which differences in web statistics are realistic and acceptable, and how the reliability of the data can be assessed for specific implementations.

In the second half of 2008, in co-operation with online marketing experts from Blue Mango, ClickValue, Maximum and Netprofiler, TNO launched a project designed to assess the measurement of - and differences in - web statistics and establish the reliability of the data. This involved analysing the data from various large websites such as Agis, Ilse Media, Typhone and Univé Verzekeringen, together with STIR and DART data.

In addition, tests were carried out on various web-statistics packages in a closed website. The packages tested were Google Analytics, Sitestat and Webtrends. For certain sites, the HBX and SpeedTrap packages were also analysed, including DART and STIR data.

Author(s):Almerima Jamakovic, Bart Gijsen and Martijn Staal