Capturing "Human Bandwidth": A Multidimensional Model for Measuring Attention on Web Sites

The diffusion and evolution of the Internet has been mirrored by an evolving science of how to measure traffic online. However, after many years of development, Web publishers and online advertisers still find the current metrics are not adequately integrated to provide a holistic picture of audience attention on the Web. This study proposes a conceptual model for measuring attention on the Web on 5 dimensions at 5 different levels of analysis: visibility (share per market), popularity (unique audience per site), loyalty (visits per person), depth (pages per visit), and stickiness (time per page). An empirical analysis of major news and information sites' traffic data identifies distinct attention patterns characterizing different types of Web sites. For example, news portals and television news sites enjoyed the most popularity. Portals and weather sites scored the highest on loyalty. Magazine sites spurred deeper visits (more pages), and weather sites outperformed other sites in terms of stickiness on a per-page basis. This model promises to be commodiously valuable for assessing a Web site's performance on all distinct dimensions of audience attention, selecting the right metrics by which to gauge performance against competitors, and by establishing benchmarks by which that performance can be quantified.