Calculator

Calculating website statistics


Many webmasters are just beginning to realize the potential of calculating statistics about website usage. Website statistics can be an invaluable source of information, revealing insights that can help you market your message more effectively. Just how many people really visit a site? Judging by the hitcount, your site might be extremely popular - but how many of these are actually unique visitors? Wouldn't it be handy to have a breakdown of the country demographics of visitors. If you're considering putting extra effort into your site, just where is the best place to direct it? Can users find the information you want them to read, or is it just buried too deep to find? Only website statistics can answer these questions.

I'm not trying to advocate any one product or service that provides website statistics. From my experience, the best way to find out what service or product will suit you best is to actually go out and try some. There's all sorts of statistic generators, ranging from graphic counters that also record IP addresses, to log analysis tools that produce classy looking reports. Once you're aware of the types of services available, and the types of statistics that are useful, make up your own mind, and see for your self by trying tools and services.

Categories of statistic generators

There are two general categories of statistic generators, those that monitor sites dynamically, and those that analyze web-server logfiles. Dynamic monitoring has the advantage that statistics are current – a website counter shows the total number of hits and is updated every time a user views the page. Log analysis generally reveals more information about visitors, and is performed at regular intervals (perhaps daily, or perhaps monthly). Both have their advantages and disadvantages, depending on the needs of the site and the webmaster.

Dynamic monitoring

Dynamic monitoring is available in a variety of forms – the most common being a counter or advertising banner that is loaded every time a new visitor looks at the site. Dynamic monitoring systems generally involve the use of an image, which is placed on either the main page of the site, or on individual pages that need to be tracked. When the image is requested by the user’s browser, a CGI script or server-side application is executed. This script or application records the IP address of the user, and then returns the image (which is often generated on-the-fly).

Dynamic monitoring is most useful for smaller sites, that don't operate their own webserver and so don't have access to server log files. There are plenty of services available, some free and some for a fee. One of the best ways is to join an advertising banner network (such as BurstMedia), and you can get hitcounts and a list of the most frequently visiting IP addresses.

Log Analysis

Log analysis provides by far the most detailed and useful statistics. Log analysis software produces reports (usually on a daily or weekly basis), which contain information about the volume of usage, the types of pages that are requested, how long visitors stay on average, and demographic information about the type of users.

Good log analysis software will produce graphs and charts that help show trends in web traffic. Some software will analyze the user-agent logs of servers to give you a breakdown on the types of browsers being used to visit (for example, percentage of Netscape versus Internet Explorer). Truly great log analysis software will examine the referrer logs of servers, to show the major external entry points to your website.

Rules of thumb

From past experience with web statistics, here are some general rules of thumb that you may find useful.

Rule 1 - Hit count isn't everything

Hit count used to be a very important measure of a site's popularity, but its usefulness is waning. There are many factors that have influenced this, including

Hit count shouldn't be used to measure your site's popularity - but it is of benefit to webmasters. It shows the rate at which people (either fifty people visiting twice, or one hundred people visiting once) visit the site. It can also be a good indication of just how many people visit individual pages - log analysis tools can rank the popularity of pages within a site, and can show problem areas such as hard to find information.

Rule 2 - Unique IP addresses aren't foolproof, but they are more accurate

Most statistic tracking services, and log analysis software, will give an indication of how many unique IP addresses visited the site. This is a far more accurate indication of the number of visitors, because a visitor will normally have the same IP address as they move through your site. When they reload a page, the count remains the same. However, they aren't foolproof, and will often under-estimate the number of visitors. Some large ISP's, or corporate networks, use a proxy server, which shows up as a single visitor even though it services multiple clients.

Rule 3 - Country demographics are very revealing

Good log analysis software offer country (and domain) demographics, that show how many visitors came from each country, and how many people come from each of the large domains (.edu, .com, .net, .mil). This information, depending on the type of site, can be invaluable, because it allows you to concentrate on weak markets, and to strengthen your hold on large markets. If you know that twenty-five percent of readers are Japanese, it might be time to consider offering Japanese translations (because if a large percentage of existing traffic speaks Japanese & English, you may be turning away an even larger number of non-English speaking Japanese). If you know that the majority of visitors come from .edu domains, perhaps its time to bring out your special "student-saver" offers, or perhaps you can market your advertising banners at a higher rate for student-related ads.

Rule 4 - Calculating visits on a per-page basis reveals a great deal

Good log analysis software will rank pages, showing the most popular pages, and showing pages that are less frequently visited. Why would this information be of interest? Well, often the less popular pages are hard to find on a site, because they're buried several layers deep. If its important information you want visitors to find, then you need to look at ways of guiding them there. If its important information, and you've had it linked to from the first page of your site, perhaps readers just aren't interested in the information, and need extra enticement. This sort of information is invaluable to webmasters.


Copyright 1998, David Reilly
Last updated : Thursday, August 12, 1999