analyze their users’ behavior. Several statistical techniques are used to analyze the data based on the measures across the four companies with seven cases. Some interesting findings indicate that certain patterns emerge according to the characteristics of the web services. Managerial implications are discussed along with suggestions for future research.
Keywords- Social Network Site; Google Analytics; Statistical Analysis.
I. INTRODUCTION
The term “Web 2.0” was first documented at a seriesof conferences [1], and its impact on the Internet has been studied widely. Many Web 2.0 collaborative tools were mentioned and three of them are particularly important, that is, Blogs, Mashups, and Wikis [2]. These tools have profoundly changed the development of internet-based social network sites (referred to as SNS throughout the remainder of the article).
Because the content of any SNS is generated by users, their behaviors are critical to the success of the sites. A study analyzed the data collected from online social networks [3]. Propagation of online social networks is another important research topic. A stochastic model was developed to explain the propagation of recommendations and cascade size [4].
Many tools are available to measure website use, and three popular Web analytics software are Google Analytics, ClickTracks, and Coremetrics [5]. Google Analytics is the only one providing free of charge, and its application in practice has been reported [6].
Because the content of any SNS is generated by users, their behaviors are critical to the success of the sites.
Working with four Web 2.0 companies, this study analyzes their Google Analytics data and aims at the following objectives. First, study the relations among them through several statistical techniques. Second, discover the commonality and differences across the four companies.
A. Case Companies and Google Analytics Measures The companies in this study are represented as A, B, C, and D with data recorded for 989, 328, 615, and 480 days during the period from years 2006 to 2009, respectively.
Company A provides a place to collect articles according to topic. Company B offers a free platform for student users, especially campus clubs, to interact on the site. Company C mainly focuses on restaurant information. Company D combines maps and blogs so that users can post their articles on the location indicating where they live.
Many measures are available on Google Analytics for retrieval. After discussion with several founders of SNSs, several important variables were identified including Bounce Rate (BR, the percentage of single-page visits or pages visited), Average Time on Site (ATOS, Average time on site. Total time on site for all visits divided by the total number of visits), and Visits (The number of times your visitors has been to your site (unique sessions initiated by all your visitors).
A. CDGR Analysis
The compound daily growth rate (CDGR) that measures the daily compound increase rate of visits from the first date to the last date of record. To have a better understanding of their visit changes over time, the CDGR measures for each case (data for company D are divided into two parts, one for members and the other one for non-members) are calculated starting from day 100 up to the final days of each case. That is, the CDGR at day 100 is the visits at day 100 against day1;
CDGR at day 101 is the visits at day 101 against day2; and so on. The equation for calculating the CDGR is as follows:
(Total visits at dayt)(1+CDGR)100=(Total visits at dayt+100)
Figure 1. Results of the continuous Compound Daily Growth Rate (CDGR) for the six cases. The X-axis does not indicate the same date for
the cases, except the bottom three charts from the same company, and it simply represents the first recorded day of CDGR.
Fig. 1 show the results of the continuous CDGR for the six cases. It is clear that company A draws a significant amount of attention at the early stage, but the pattern shows that the curve has a negative slope in the long run. From the longitudinal observation, Company B undertook several large-scale promotional campaigns and they are revealed by the up-and-down shape of the curve. In terms of stability, company C has the best performance with very few fluctuations. In addition, a large portion of the curve remains above zero, indicating the steady increase of its users. Although company D records the largest number of visits and registered members among the four companies, its curve shows that the number of its visitors fluctuates moderately in which the curves of members and non-members reveal an interesting pattern. A gap of approximately 50 days appears between the two curves. Do member users follow the behavior of non-member users, or vice versa? What causes the time lag? These questions remain unanswered pending further investigation. To detect abnormal data, Multiple regression models with a 50 days of moving window is also used and results show that each case behaved differently with certain patterns emerged.
B. Time Series Plots for %NV
%NV is an indicator of how the site attracts new users, and is expected to decline over time, since the base of new users is shrinking. Fig. 2 shows the %NV of companies A, B, C, and D over recorded days. As predicted, they all decrease as the longer they stay in business. In general, company B is in a desperate position with its low %NV. The curves of companies A, C, and D stabilize at approximately 50%. This can be used as an indicator of a stabilized SNS.
Companies A, B, and D seem to experience a major drop before climbing to a higher %NV. Is this an unavoidable obstacle any interaction-oriented SNS is forced to experience and overcome, or simply a coincidence among multiple web services. Statistical techniques were applied to study the data, and the results show some interesting findings. Further study is suggested to provide a better understanding for the users’ behavior on SNS.
ACKNOWLEDGMENT Management Journal, vol. 41, Jul./Aug. 2007, pp. 25-33.
[3] V., Viswanath, A., Mislove, M., Cha, & K.P., Gummadi, “On the evolution of user interaction in Facebook”, Proceedings of the 2nd ACM SIGCOMM Workshop on Social Networks, Barcelona, Spain, August 17, 2009.
[4] J., Leskovec, L.A., Adamic, & B.A. Huberman, “The dynamic of viral marketing”, Proceedings 7th ACM Conference on Electronic Commerce (EC-2006), Ann Arbor, MI, USA, June 11-15, 2006.
[5] M. Chafkin, “ Analyze this, says Google,” Inc., vol. 28, Apr. 2006, pp. 30-30.
[6] L. M. Braender, C. M. Kapp, & J. Years, “Using Web technology to teach students about their digital world,” Journal of Information Systems Education, vol. 20, Summer 2009, pp. 145-153.
-0.02
無研發成果推廣資料