Making Sense of Web Site Traffic

By Baselinemag  |  Posted 2002-10-01 Email Print this article Print
 
 
 
 
 
 
 

In trying to make sense of Web traffic, many managers get distracted by the wrong numbers or frustrated that the numbers they think they need aren't available. Here's how to avoid both pitfalls.

Your Web server's traffic logs can tell you if your site's page hits are going up or down, but how do you know what visitors are really doing at your site?

PDF DownloadEnter Web analytics, which aims to extend site traffic measurements by tracking and reporting visitors' behavior when they click into, interact with and leave a Web site. Ultimately, the insights you glean here can help you prioritize site improvements.

How finely your data can be analyzed may depend greatly on how your site handles personalization and secure transactions, according to independent consultant Jim Novo. "On the content side, the site may not be able to distinguish a unique visitor," says Novo. This may happen, for example, when Web sites balance their workload across multiple servers during times of high traffic. A single visitor session may be handled by several different Web servers. As a result, log files from multiple servers need to be correlated in order to capture each visitor's experience.

A June report on Web analytics from the Yankee Group notes that retail sites and large-scale Web sites derive the greatest benefit from deep reporting on customer behavior. These analytics products provide data based on such criteria as number of pages viewed, time spent at the site, whether the visit is by a return user, and exit point.

Novo points out several issues to keep in mind to ensure that you use Web analytics consistently and effectively:

  • Standardize how you define Web metrics within your organization. If you change the definition of a term like "page view" over time, for example, it will be difficult to compare new metrics with previously gathered data.

  • Don't despair if it's not possible to get an accurate count of a particular activity. Try a different perspective, such as looking for a trend. In the end, you may find that an absolute statistic ("We received 44,963 visitors last week") may not be as meaningful as the relative number ("Visitors dropped by 10% last week").

  • Web analysis should also be anchored within a timeframe, since its real value is in how the results change over time. To calculate sales per visit, for example, you should calculate your sales for a specific time period as well as the site visits for the identical period. Always use equal time periods when creating ratios, percentages and indexes.
    •  
     
     
     
     
     
     
     
     
    Manage your Newsletters: Login   Register My Newsletters