If you’re a long-time Getting Informed reader, you will remember multiple previous references to the importance of web statistics, or stats, in understanding the effectiveness of your web site. In the past, most web site owners were just happy to see that cheesy hit counter increasing…you know, the one that looked like a digital readout. Today, it seems that the statistics being provided are bigger and better—and quite overwhelming for most site owners. And just to confuse things, the term analytics comes along to see if you’re paying attention. Analytics are statistics in a way that speaks to the average person rather than to the classic statistician. In short, analytics are where the rubber meets the road. But for the purposes of this discussion, information about site viewing data will be called statistics.

The two biggest problems I find in discussing site statistics with many business owners are these: 1) the site owner doesn’t know how to get to their statistics; and 2) the site owner doesn’t understand how to translate the numbers into information they can use. This month’s column will be dedicated to responding to these issues. 

What are web statistics?
Web statistics are simply an analysis of internet-based requests to your web site, or traffic. Statistics packages should provide a simple-to-read interpretation and display of log files generated based upon traffic or by tags embedded in the pages of your site. Log files are typically generated using Common Logfile Format (with a Linux-based Apache web server) and, once enabled, can be analyzed by a variety of statistics packages like AWStats or Webalizer. One of the best tagging utilities available today is Google Analytics. Other packages use a combination of reading log files and tagging.

Web statistics are configurable and the configuration can have a major impact on the final numbers. This is especially true when it comes to time frames and what is a legitimate page requester. With statistical configuration, accuracy should be the most important factor. Over-inflating these numbers does nothing to help you. Over-inflation can occur when time periods are too short or hits are counted that should be omitted (such as your own hits or those of robots or spiders). If you get sucked into a “my site gets more hits than your site” conversation, focus on what your site does well and ask, “But what is your bounce rate?”

Where are my web statistics?
The location of your web stats is dependent upon the statistics package used, your hosting package, and selections made by your designer/webmaster. If the most popular packages have been used, your statistics will be password protected and available through a particular directory on your site. Frequently, stats can be available at www.yourdomain.com/stats or /webstats. Sometimes your web stats are available only through the control panel or administrator utility of your web site. You may also have a link on your web site that provides access to your statistics.

If none of these apply, ask your webmaster or designer for the location of your web statistics. If you manage your own site, you can check with your hosting company to find out if a statistics package is installed and, if so, how to gain access.

What do these numbers mean?
The names and figures that appear in your stats are dependent upon the package you use; however, there are some basic numbers that give you the most important information, so we’ll focus on those. I’m also including some that provide traps—statistics that are currently cited but that don’t provide as much real value as it might seem.

How do these numbers make my site better?
Statistics in and of themselves mean nothing—like most historical data, they are important only as a tool to make future improvements. Also, it is hard to make generalities about your statistics without looking at your specific site or looking at statistics in tandem. Like so many things in this business, it all depends. For example, when evaluating a high bounce rate, it’s important to see what the viewer is seeing when they come to your site—what is the look and feel; how long does it take to load the page; is the navigation evident; and is the content compelling?

When attempting to improve specific statistics, there should be a clear understanding of the business purpose behind the change. In the example above, you could assume that a high bounce rate means a loss of potential clients or leads—a bounce means that the user has not continued to investigate your organization beyond the page through which they entered your site. If they don’t click anything else on your site, they’re definitely not buying anything or submitting a request for contact, so they probably qualify for a missed opportunity. On the other hand, a high bounce rate may be a result of your site attracting other than a legitimate client. If you have a highly targeted market coming to your site, the bounce rate will be minimal…as long as you have an attractive site that loads quickly with easy navigation and applicable, interesting content. See how it gets complicated? Identify the most likely culprit and resolve that item first. Sometimes to evaluate your site fairly, you will need to involve an external party. After making the first planned change and allowing sufficient time to pass, evaluate whether the change brought about the desired results. Although an improvement in the statistical measure is the first view of improvements, the ultimate measuring stick for success is the same as it has always been—are sales increased; are costs reduced; or is profit margin improved?

Obviously, this approach takes time and finesse. It’s also an area that requires a planned, phased approach. The expectation for results must be realistic and time must be given to really see a true picture. These concepts are on the other end of the traditional spectrum for the Internet—I want it, and I want it now.

What’s the best statistics package to use?
There is a lot of debate over the best statistical package, but in my opinion, the best package is the one that the site owner can review and really understand. That means you should take a look at a couple of different packages to see which one you find easiest to understand. Many very good packages are Open Source software and free to the user.

Of late, the favorite among our customer base is Google Analytics (GA), which is an adaptation of Urchin.

After a week of having access to Google Analytics data, one of our clients told me it dramatically reduced their time to create an advertising budget. I must admit that the interface is one of the simplest to understand. Most site owners are especially impressed with the global map that allows drill-down to see the number of hits per city, called GeoTargeting—pretty impressive data that would take significant time to assimilate the old-fashioned way. Another nice feature within Google Analytics is the integration of AdWords tracking, other campaigning efforts, and statistical goals.

Google also manages to counter one of the ongoing criticisms of statistical packages—a lack of long-term historical data. Many users have a difficult time figuring out if their redesigned site is as good as their old site because their old data is gone with each hosting change. Since GA doesn’t reside on the site itself, stats are still at Google regardless of where the site is hosted. Google also allows any and all reports to be exported to PDF and, depending upon the report, to CSV, a file type easily viewable in Microsoft Excel. This feature puts the power of record-keeping under the site owner’s control, even though Google commits to maintain the data long-term.

And did I mention that Google Analytics is free? Although critics of Google accuse them of “Big Brother” tactics by offering such a tool, it’s obvious that something that makes possible the understanding of traffic data for millions of sites is an exceptionally valuable tool. 

No matter what statistics or analytics tool you use, the important thing is to have statistical data available to you, check it on a regular basis, and understand what you’re looking at. For additional information, check the following resources regarding statistical analysis: en.wikipedia.org/wiki/Web_analytics, www.google.com/analytics, and usability.about.com/od/usabilitytesting/a/measusability.htm. 

This article used with permission from 400 Edition, Inc. It is unlawful to use this article or any portion of it without the expressed permission of its author. This email address is being protected from spambots. You need JavaScript enabled to view it.