Using Web Analytics to Investigate the Navigational Behavior of Users

Abstract Today, web analytics and big data provide the basis for the evaluations of websites. However, the interpretation of this data in terms of Human Computer Interaction (HCI) can be challenging and there has been little progress in the systematic exploration of the potentials of web analytics within the usability evaluation domain. In this paper, we introduce a new approach to analyzing large amounts of usage data with the goal to detect usability problems in the use of a website. The underlying idea was that the analysis of patterns can support the HCI analysis of a website. To this end, we developed an extension for the open source web analytics package Piwik that allows investigating the navigational behavior of users through the abstraction of individual interaction paths. The developed system supports data mining on navigation logs and the discovery of interesting patterns. 1 Introduction Today’s relevance of the internet for private and commercial use can hardly be overestimated. Today, more than 50 million people in Germany use the Internet as integral part of their daily lives (Bundesverband Digitale Wirtschaft 2013). This includes all age groups, whereas 14-29 year old Germans have the highest online penetration rate. According to Bundesverband Digitale Wirtschaft (2013) almost all internet users (97.1 percent) have already searched for information on a product or service on the internet. However, offering a web page alone is not enough to guarantee commercial success. Along with design and content the navigation of the web page plays an important role. The best way to investigate the usability of a website is to conduct a dedicated user test in order to understand how potential users cope with the navigation. Unfortunately, running such tests is associated with high costs which in particular small and medium-sized enterprises are not able or willing to take. Log files are a comparatively cheap alternative to access large amounts of usage data. In this paper we introduce an approach for the investigation of the navigational behavior of users that combines web analytics with knowledge about the structure of the website. The main goal was to take the navigational behavior of real users and to determine patterns that signify usability problems (e.g. disorientation on a web page). We wanted to use the structural information about the web page to supplement the log file analysis. This means linking the information of the log files to the structure of the page. In order to allow for an