Unveiling the Traffic Dynamics of A Virtual Community

The research results of a quasi-experiment on an award-winning website are reported. A virtual community was built on this website to study its performance under the influence of two treatments. The first treatment was an administrative modification – a change of membership policy that restricted nonmembers’ access to a popular area of the website. The second treatment was the addition of an advanced feature that helped cultivate a virtual community, using combined push and pull capabilities. Data of four traffic performance metrics were recorded: pageview, visitor count, daily registration, and average duration. The data points preceding the treatment were analyzed using time-series ARIMA model to predict the data points subsequent to the treatment. In order to illustrate the effect of the treatment, the actual data prior to the treatment, the actual data following the treatment, and the predicted data are plotted on the same graph. This allows a clear visual inspection of how the treatment affects the data trend. The results show that for the first treatment, pageview, visitor count, and daily registration, all increased considerably. However, the levels of pageview and visitor count dropped back down very fast, while daily registration stayed high. Conversely, the seemingly positive treatment did not result in an increase of average duration. Unlike other performance metrics, average duration decreased. For the second treatment, the levels of pageview, visitor count, and average duration not only increased, but also stayed high. However, the treatment did not tend to change daily registration, as the level did not change much.