Picture the scene: A marketing manager spends thousands of dollars planning and executing a high-profile integrated campaign and then … does nothing. No reports on its success or failure, its visibility, or the demographics of the audience reached.
Sounds incredible? When it comes to Web sites, this scenario plays out time and again.
The more you know about the strengths and weaknesses of your site, the better it becomes. Just about every site has its hit-counter, but it is necessary to go much deeper to make the most of your company's Web presence. Critical e-business intelligence must include visits to each page, visitor domains, average time spent viewing pages, whether changes increase or decrease traffic, which advertising vehicles are most effective, busiest times, paths visitors take and how many visits resulted in sales.
Putting this information to use is easy. A common path taken by visitors is from the home page to the general product information page and then to a specific product's page. This can take four or five clicks on some sites, and only one or two on others. Making it easier for people to get the information they need encourages return visits. This is a technique called designing for analysis. That is, the site should have a logical structure which enables you to track visitors meaningfully.
Another example of how analysis can deliver benefits is in the capture of a visitor's Internet domain. With these addresses, you can build a database of visitors, gaining critical e-business intelligence about a given Web audience.
Harvesting the Data
So far so good, but how does one go about gathering this data and using it?
Some of the information can be gleaned from the Web server's log files (files that capture each individual interaction between a Web site and a visitor). However, these logs were not designed with in-depth site analysis in mind, and cannot easily be tweaked to produce desirable marketing information. The challenge for e-business analysis software is to take the log file data and turn it into useful information about users and their visits.
Several factors can make it difficult to track visitors.
IP addresses, for example, are a way to identify unique visitors. However, many Internet service providers employ dynamic IP addressing, a method of spreading a large user demand for IP addresses across few machines. As a result, a visitor may come from different addresses on different days.
Proxy servers, applications that break the connection between the sender and the receiver, prevent an outside source from receiving details of a private network. All Internet requests from inside a security firewall must first go through the proxy server. In large companies, hundreds of PCs may make requests to a site, but only the host name of the proxy server will appear in the log. Also, frequently requested documents are stored locally or cached on the proxy machine. When visitors request such documents, the proxy just returns the copy it has on disk and no request is made to the original Web server.
Each HTML page requested is logged by a Web server independently — without describing how one is related to another-making it difficult to track people as they move through a site. If traffic is high, it becomes difficult to piece together a continuous visit from one person. As a result, an e-business analysis solution must be smart enough to thread page requests across multiple machines.
One answer to tracking difficulties is cookies, identifiers used to store information about a user's interaction with a site. When a cookie is correlated with online behavioral information, a marketer has the ability to understand concepts such as visit recency and frequency.
The Need for Analysis Tools
But cookies are not a complete solution. Given the importance of tracking, software allowing on-site analysis is essential.
The analysis software should use an online database so the marketer can use historical data to look for trends. It should also package data in a variety of report formats, from raw statistics to presentation quality. That way, CIOs, e-business managers and Web site administrators can decide whether to provide basic figures, for say, IT analysts, or charts and tables for marketing-department presentations. Various software formats can also be specified for the report such as HTML, Postscript or PDF.
The software should also offer the ability to run multiple parallel reports and import and analyze other Internet server log files. It also should allow production of these reports to be automated so the e-business manager has the information he needs, as soon as it's available.
As Web sites grow ever more central to successful business, analysis of the way they are used becomes a fundamental issue. An effective tracking package is a critical part of ensuring Web site effectiveness and measuring return on investment.
Larry Bohn is the president/CEO of net.Genesis Corp. Reach him at [email protected]