What is a Visitor Anyway?    1    2    3    4

Part II of IV: What is a Visitor Anyway?

By Dane Christensen

In Part I of this series, we discussed the concept of visitor sessions being defined in terms of the amount of time between page views. But wait—how do we even know that the page request was made by the same person who requested the other page six minutes ago? Well, that depends on whether you're using session cookies or heuristics to identify visitors.

Yikes! What!?

Not to worry. Each of these concepts is pretty obvious once you understand them--we'll take them one at a time. Let's tackle the term "heuristics" first.

Heuristics: A 'Reasonable' Assumption
One is said to be using "heuristics" when he or she uses previously acquired knowledge to solve a problem. So when it comes to determining visitor sessions, a heuristic approach would use data from previous requests to determine if a current request is part of a visitor session.

Consider this: A typical log file request contains several pieces of data that can be used to identify visitors and actually piece together a visitor session. For example, the log file shows their IP address and the type of user agent (i.e. web browser) they're using. We can also see the time that they request a file and the page they were on just before they clicked to the current page (i.e. the referrer).

As ClickTracks examines a log file, if it finds a page request where the user agent and IP address match those in a previous request—and, if that occurred within the 15 minute session timeframe, and the previously requested page is the same as the referrer for this request-- then the program can be reasonably sure that the same person made both requests and should be counted as part of the same session

To make it crystal clear, here's an example: If ClickTracks sees two requests from a Mozilla Firefox v. 5.2 Windows browser, and IP address 178.200.17.1, both made within 15 minutes of each other-it is reasonably sure that both requests are from the same visitor and includes them both in one visitor session.

Defining 'Reasonably Sure'
We say "reasonably" sure because there are scenarios that can muddy the picture somewhat. For example, if multiple people from the same corporate office both go to your site, chances are they both have the same IP address and user agent. Also, people logging into the Internet from dial-up accounts with the same ISP may be sharing IP addresses and coincidentally have the same user agent. And then there's good old AOL, with millions of people hitting the web through their proxy servers, (same IP addresses) all using AOL's browser. Situations like this can cause two or more different visitors to blur into one. So while determining sessions using heuristics is fairly reliable, it is by no means perfectly accurate.

Cookies-Another Way to Define a Session
With session cookies, when a visitor first visits a page at a particular time, they're assigned a unique ID which is written into a cookie. On their subsequent page views, that cookie is recorded in the log file. If they have been inactive on the site for a specified length of time (e.g. 15 minutes), then the cookie will automatically disappear, and their next page view will be the start of a new session. This eliminates the need to rely on heuristics.

Now ClickTracks only needs to find the requests with matching session IDs to be absolutely certain that these requests are part of the same session. Of course, there is the issue with people who refuse to accept cookies, but the accuracy gained by tracking with a session cookie more than offsets any lack of data due to these rare cases.

Session Cookie Setup
Tracking visitors by session cookies requires a bit of additional setup on your part. First, the cookies must be checked and/or set by your web site--and this must happen consistently across all your pages. Second, your web server must be configured to record cookies. If you haven't specifically set it to do so, chances are it isn't doing it now. Last, you must identify the session cookie within ClickTracks, specifically within the Dataset properties on the ClickTracks Pro Server. (Cookie tracking isn't available on Analyzer or Optimizer, so if those are the programs you use, you'll have to rely on heuristics— or you can always upgrade!)

Another thing to remember about session cookies is that the session timeout setting (e.g. how much time elapses between page views before it's counted as a 'new' visit) is no longer established within ClickTracks' settings, but by the cookie itself. That is, the cookie is set with a certain time to live. If another page is viewed within that timeframe, the cookie is reset and the session continues. If that timeframe elapses, then the cookie will expire and the next page view by that person will cause the creation of a new cookie with a different ID, and therefore, a new visitor session. While identifying visitors with cookies will give you more accurate data, it also may require some coordination between your marketing and web development staff.

Generally, the differences in your reporting will not be radically different between these two "sessionization" methods. However, that can vary depending on the nature of your visitors. Do an inordinate number of them come from corporations, certain foreign countries, or AOL? If so, the improved accuracy of cookie tracking may be worth the extra effort.

continued >>

What is a Visitor Anyway?    1    2    3    4