Understanding user visiting behavior and web design : Applying simultaneous choice model to content arrangement

A common problem encountered in web design is how to arrange content on the homepage of a website. This paper uses a random-utility theory in studying visitors’ choice behaviors to optimize web design. Classical discrete choice models are not suitable. A total of six multiple-choice demand models are proposed in this paper. These models are applied to web log file data collected from an educational institute over a seven and a half month period, and the parameters are estimated consistently across all models. The best model based on the forecasting accuracy rate is selected as the tool for resolving the problem of web design. Two metrics, utility loss and compensating time, are constructed using the selected utility model to facilitate web design. Empirical results show that the proposed metrics are highly efficient to develop web design to resolve the problem of how to allocate the information resources of a website, and the algorithms can also be utilized to assist the study of the feasibility of introducing a new function in a website.