IT Werkz Sometimes

How come site visit statistics from different counters never show the same number of visits?

And they never do. You could put six counters on a site and within days the numbers of visits will not be the same. What’s worse is that quite often the counter writers try and justify it. I just came across the following at Google Analytics:

Why does Google Analytics report different values than some other web analytics solutions?
Different web analytics products may use a variety of methods to track visits to your website.

Do they? Why? How can a counter have a different method of tracking visits to a web site? If I start a new blog with 6 different counters on it and no visits i.e. the current hit count total is zero, and then I go to 6 different PC’s and visit the site once from each machine then that site visit total must be 6. And if I continue doing this kind of test in a controlled way, so that I know exactly how many visits I have made to that site then the displayed count totals for all of the counters have to be the expected result total. Period. More from Google:

Therefore, it is normal to see discrepancies between reports created by various products. However, we generally believe that the best way to think of metrics across different web analytics programs is to think in terms of trends, as opposed to numbers by themselves.

Well that would be convenient wouldn’t it. Never mind the numbers, forget about them coz we just can’t add up, think of trends instead. Of course the problem is not that they can’t add up its that they lose the information somewhere. I’m getting a right bee in my bonnet re lousy stats.

I ran the test yesterday where I visited my own blog’s pages at specific times of the day, which were noted, so as to check that all visits were recorded on the log file.

And were they? Yup. Everything showed up like where I expected them to. So confidence has been restored with regard to the raw log file. Now I’ve just got to find some analysis software that will read the log file and present the data in a more user friendly fashion rather than the ASCI file dump I currently see. The only way this is going to work is if the third party analysing software gets access to the log files which means I have to give them the FTP username and password. Is there going to be a problem with that? Wouldn’t have thought so as I only have read access so no damage can be done. This sounds like something I get the opportunity to pay for.

A couple of days ago I signed up with Google Analyitics to see what kind of statistics they knocked out. The setting up of these stats said something like add some javascript to each page I wanted analysing. Do they mean page or blog entry? Dunno. Anyway I added the javascript to the two most recent blog entries to see what I got out of that. And what have I got out of that. Haven’t a clue.

It seems that Google Analytics is analysing more than just the two pages with the javascript code as the stats include other pages. Of course the daily totals don’t map onto the raw log file totals. Nor do they bear much resemblance to the Sitemeter stats or the Blogrankers site or even the Blogtopsites or Blogtoplist sites. In fact none of them have daily visit totals or even total site visit totals that resemble each other in any way shape or form. Nielsen it ain’t.

I’m thinking of going to the Starwest Testing Conference in Anaheim in October. One of the speakers is Jon Bach who is the Exploratory Testing guru. That should be interesting.