Leveraging Web Data for Competitive Advantage
Can companies leverage Web-traffic data for a competitive advantage? The answer is "yes."
First, some background information is required. Any analysis of Web traffic requires data that is created by the Web server software. This data contains Web log records known as hits. Every item on a Web page generates a hit. Logically, the more pictures a Web site owner puts on a page, the more hits the Web server records in the log. Page views are simply the number of times Web pages have been seen by a user. Visits are the series of pages seen when a user comes to (visits) a Web site. Typically, there are multiple hits per page view and multiple page views per visit. From our analysis at IBM, we see an average of five hits per page view and typically an average of five page views per visit.
What is important to measure and what is not? There is a maturity cycle through which all Web-site owners progress as they experience the power of the Internet. As a Web site becomes an integral and strategic part of a company's business plan, Web-site owners want more information. They realize that reporting on only hits or even page views, while interesting, doesn't provide an ability to improve site effectiveness, particularly when this statistic can be easily manipulated by Web site design.
However, investigating and interpreting visit patterns can be very insightful. Looking at the most prevalent paths site visitors travel, where they came from and how long they spend on specific pages or categories of pages are important techniques in measuring a site's success. Investigating content affinities and behavior patterns of site visitors can help determine the most effective placement of information. In addition, understanding the quantity and quality of visitors, for example, understanding new visitor behavior versus repeat visitor behavior, can make a dramatic difference in a Web site's success. With this analytic capability, companies gain meaningful insight into their users and their users' behavior allowing them to target their content. They gain improved customer satisfaction and, ultimately increased sales.
The three elements of Web-site analysis that prove to be most important are path analysis, referral analysis and the use of advanced data mining. Any such analysis, to be effective, requires a well designed, underlying database and a fail-safe method of populating that database with clean data. This data must be stored in a manner that the sequence of pages each site visitor sees is when they come to a Web site (visits) can be determined.
Path analysis is a methodology of investigating what pages are seen before and after a specific page. This analysis includes determining time spent on each page as well as entry and exit points from the Web site. It is an effective technique for measuring content affinities as well as effectiveness of site design. It can also be used to measure the impact of creatives such as animated pictures, added to a specific page with the intent of attracting site visitors to a specific part of the Web site. The Internet is a wonderful medium for experimentation with marketing and design strategies, for one can change content, immediately measure the results and take appropriate action based upon those results.
Referral analysis is also very valuable. This process determines where Web site visitors came from before visiting a Web site. It can be very effective in measuring the "clicks" on ads placed on other Web sites and the subsequent Web pages these "clickers" investigated. Such analysis can also reveal how long they spent on each page and the duration of the entire visit. This information can help assess the value of affiliations with other Web sites or search engine traffic, including search terms used that resulted in those valuable visits to the site. It helps answer the question, "Where do the best customers come from, what are they looking for and what do they do when they get there?" This analysis allows the Web site owner to make appropriate modifications and to structure business agreements based upon fact and not upon intuition.
Data mining can be an ambiguous term when used indiscriminately by site analysis tool vendors as well as Web site owners. Many will use the term when they have implemented only a query interface for analyzing data. This application is referred to as verification-driven data mining, where the user validates a hypothesis such as," traffic from a specific location is increasing at a x% rate." A more sophisticated implementation of data mining is known as discovery-driven data mining and uses techniques such as clustering, associations and neural induction. This application requires advanced software and yields results created by the data relationships and not by a supposed hypotheses. This advanced data mining technology can be very valuable in determining content affinities as well as site visitor behaviors, which can lead to increased content effectiveness.
Path analysis, referral analysis and data mining can be effectively implemented without the use of technology that is perceived to invade consumer privacy, such as registration or the use of persistent cookies. However, if either technology is implemented, a significantly more robust analysis can occur, showing for example, behavior comparisons of new visitors versus returning visitors or even analyzing demographic patterns by visitor behaviors and content areas, such as, "show me the incomes and age levels of visitors to specific pages or content areas on our Web site."
For the Web-site owner who truly believes in the power of the Internet to effectively deliver solutions, the time and resource spent in analyzing Web traffic can yield significant competitive advantages. It does require an investment of time and talent to determine those valuable "ahas" that will deliver a competitive advantage in this new worldwide medium, but it is an investment with a great return.
John Payne is solutions executive at the SurfAid Analytics unit of IBM Corp., Armonk, NY. His e-mail address is email@example.com.