Your Internet Statistics Can Lie

Performance metrics are of capital importance for all managers who wish to calculate the profitability of an e-business activity, marketing campaign, public relations operation or any other business investment [1] . Many of our previous articles have commended the quality of Internet metrics and have stressed how important this data is for improving both your online and offline performance. While this is certainly true, it is important to put the precision of some of this data into perspective. Sometimes, these stats can paint a picture that is very different from reality.

For this reason, it is important to analyze your data carefully if you want to avoid undue elation or panic when reading your stats. Remember that statistics are indicators, but they are never 100% reliable. Here are a few tests for verifying the accuracy of stats that track your number of visitors and the length of their visits.

100,000 unique visitors saw my site last month…

Are you sure about that? There are many factors that may falsify the statistics provided by metrics tools:

For instance, if the clientele of your site is composed mainly of business customers, you are likely underestimating the number of unique visitors to your site. This phenomenon is due primarily to the fact that while most tools use IP addresses to identify visitors, many large corporations use only one IP address for all their computers. As a result, your statistics could lead you to believe that a single person visits your site very frequently when in fact there may be 1,000 different people visiting it from the same company or government department.

On the other hand, if your traffic is comprised in large part of individuals, the opposite can occur. Most Internet providers assign different IP addresses to individuals each time they connect to the Internet. Consequently, you may come to the conclusion that 10 separate people visited your site when in reality the same person returned 10 times.

Visitors spend an average of 10 minutes and 32 seconds on my site…

How is that number calculated? More often than not, this stat is calculated by comparing the time a person takes to click between one page and another on a site. In other words, a person who visits a page on your site and clicks a link to open a new window 5 minutes later will be counted as a five-minute visit. It is therefore impossible to know how much time the visitor truly spent on the last page he/she viewed. For the same reason, a person who visits only a single web page will be calculated as a zero-minute visit because there is no basis for comparison. Depending on how your tools analyze this data, the average time spent on a site may actually be a very unreliable statistic.

Solutions…

Cookies can be used to reduce visitor identification errors by identifying the person rather than the IP address. However, some users block cookies or regularly delete them, meaning that even cookies are not entirely reliable.

Independent solutions installed directly on your site can also improve the quality of your stats. Such solutions are particularly useful for reducing the falsifying effects of pages stocked in cache memories, robot visits or pages that are not completely downloaded. These alternative solutions are in fact not widely known by the public.

The best solution still remains to analyze your statistics by looking at them as a whole rather than as individual numbers. By comparing stats on a month-to-month basis, you will be able to see if your situation is improving or deteriorating. If you would like to make precise profitability calculations based on your site’s statistics, a statistical audit would be your best solution.

[1] To find out more about performance metrics on your Web site, consult the article Understanding Web Performance Metrics.