by Marc Bissonnette

Everyone knows (or should know) what web traffic statistics are, at least in the basic sense. What most people do not know however, is all the information they contain (actually, that should read "all the information they can contain", depending on your ISP).

First of all, my caveat: My first article with Net World covered the fact that not all stats are necessarily accurate or complete, due to the large amount of caching servers on the Net. With that being said...

There are far too many site owners who use their stats solely for the purpose of determining total number of visitors to their site. While this is nice to know, it certainly doesn't help you determine which areas of your site need more work, which are the most popular, which might be attractive to advertisers, and which sections can be discarded altogether.

So just how do your stats do tell you all of this? First of all, you'll need one or more of a few things;

• A stats analyzer (or one being run on/by your ISP)
  
  

• Patience

Now, having access to all your server logs isn't necessarily a guarantee that you'll get all this wonderful information. You have to make sure the ISP is tracking all the data you need. Some things like http_referrer and user_agent are left out of server logs to save space. (http_referrer tracks which page linked to yours, pages visited, the path users took through the site, average length of visit etc., and user_agent tracks which browser people are using to view your site, which Operating System they're using, etc.). If your ISP doesn't track these right off the mark, you can ask your ISP to implement them either in general or for your particular VWS (Virtual Web Server).

So, assuming you have access to this kind of data, how do you access it all? We use MarketWave's Hitlist Standard (freeware version, although there are some very impressive "pro" versions in the US$295 to US$995 range) for our statistics analysis.

There are many other excellent statistics analyzers, running on all platforms, as well as versions that will run directly off of your server.

So what do you do do with all of these numbers?

There are a zillion different uses for these numbers and I'll try to cover a few of the more important ones here. (The section titles are used from HitList Standard, your stats analyzer may have different titles, but they're fairly self-explanatory)

Total Visitors / Requests

There is an important difference between the two (Visitors and Requests). Total Visitors is the total number of unique visitors to your site. Total requests are the total files transferred from your site. (Remember, one visitor can make 15 requests in a single visit: Consider: One page, with fifteen images and one java applet, is 17 requests or hits: 1 for the HTML page, 15 for the images and one for the Java applet). I've seen a lot of people get really excited by seeing 15,000 requests to their site, but then come crashing down when they discover that those hits all come from 300 people alone.

Most Popular Pages

These, as the name implies, are the most frequently viewed files within your site. What this should mean to you, however, is that these are the pages where you will get the most bang for your buck, so to speak. These are also the pages that you would use to drive traffic to other, less used sections of your site that you wish to draw attention to. These numbers are also the ones the advertisers will want to see to ensure that they're not being placed in the "back corner" of your website.

Most Popular Entry Pages

These are the pages where people are first coming into your site. This can occur either from the users typing in the URL directly, from a bookmark file, a search engine or a cross-link from another site (Remember how important cross-linking is!)

Most Popular Exit Pages

These are the pages that users last viewed before leaving your site. These are a good indication of which pages may need some tweaking in order to keep people within your site.

Most Popular Single-Visit Pages

These are the files that people are entering your site to view and leaving without visiting elsewhere. These are also good targets for tweaking to encourage users to visit further within the site. Chances are that the users have viewed the site and left immediately for one of the following reasons:

1. the page they entered did not appeal enough to them to continue to visit

2. wrong number :-)

3. the information they were looking for was not clearly defined (similar to 1) above)

• b) she is typical of the target markets

• c) she was not involved in the process of design and therefore is less biased

• d) she loves finding faults :)

This is usually the list of countries people are visiting from. It's amazing the number of companies / business owners that discount international visitors to their sites. Unless you're selling groceries in a corner store, chances are you have the capability of international business of some nature or not. Consider your home nation's dollar, as well. Personally, as a Canadian, our weak dollar versus the American dollar makes InternAlysis very attractive to US firms looking for our type of services. I would imagine that a great deal of the readers of this list are from New Zealand, which is in an equally attractive position for international marketing.

As an example, my own site has traffic from United States, Canada, United Kingdom, Denmark, Japan, Brazil, Australia, Trinidad and Tobago, Norway, Netherlands, New Zealand, Israel, Malaysia, France, Saudi Arabia, Switzerland and Singapore.

If you think that because my business is Canadian based, I ignore requests for service from all of the above, you'd be quite wrong! :)

Depending on the types of services and / or products you offer, you may want to consider keeping your international visitors in mind when updating your site (For example, I'll be doing something along these lines next week with internalysis.com)

Well, those are the most common logfile elements, though by no means are they the only ones. Depending on your log analyzer, you can manipulate your display to show literally hundreds of combinations of the data and their patterns/trends

I should also mention that for those of you who have further questions or comments about this article, or any of the others I've written, please feel free to email me at dragnet@internalysis.com.