Adjusting Cross-Domain Analytics Data

Anyone that’s used Google Analytics to track cross-domain requests has probably run up against the fact that Analytics adds some really ugly GET variables to the end of your URLs when you click on links. Not only are they ugly, but they also can stop things like WP Super Cache from caching your pages. We also found that the query string appended by Google Analytics was causing server errors when appended to the URLs of some of our hosted apps.

There is a little-publicized feature in Analytics, though, that lets you change the query string into a hash string. Therefore, instead of having some long, ugly string that can mess things up (and, to be honest, long, confusing query strings can sometimes scare users); you get a long, ugly hash appended to the URL, instead (which has no effect on the way the page is rendered, and, therefore, doesn’t mess up nearly as many things).

Firefox and HTML5

Earlier this evening, I was reminded of just how primitive the HTML5 support was in Firefox 3.5/3.6. While we have seen three major version releases since 3.5, it was actually still the latest version of Firefox less than 6 months ago (and was that way for almost 2 years). Therefore, Firefox 3.5.x still holds a decent amount of market share (probably as much as, if not more than IE6 did a year or two ago). Looking at a handful of websites for which I have analytics data, versions of Firefox prior to 4 still accounted for anywhere between 2% and 15% of the total visits to those sites last month.¬†With all of that information, it’s probably still important to make sure your sites work in versions of Firefox as far back as 3.5.

There are two somewhat major gotchas in the way Firefox 3.x handled HTML5. The first is easily fixed with a few lines of CSS. The second can only really be fixed if you rewrite some of your HTML.

Track AJAX Requests with Analytics

Google Analytics has included the ability to track AJAX requests for quite a while, but I don’t know how much awareness there is of this particular feature. Personally, I just discovered it the other day. I was looking at my Analytics data for a particular page on which I heavily use AJAX to replace all of the body content based on form selections. At that point, it occurred to me that, while I was getting good information about the numbers and types of people visiting the page, I wasn’t getting any segmented data based on the information they were requesting.

Woopra Going Public

Woopra, the real-time Web statistics tracking application, has announced that it will be moving out of the private beta phase very soon. When it moves out of beta, Woopra will be introducing a “freemium” business model, allowing most non-commercial users to use the service for free. However, those free users will be limited to 30,000 pageviews per month (not a big deal for most non-commercial sites, I’m guessing) and will only be able to store 90 days of historical data.

I have to wonder, however, how popular the commercial model will be for Woopra. With so many other groups out there offering free stat tracking for Web sites, with little to no restrictions, will the real-time information and desktop application be enough to get people to pay for Woopra? I suppose only time will tell.

Online Audiences and the Paradox of Web Traffic

If you are as much of an analytics nut as I am, then the video below is for you. It features Dr Matthew Hindman, Assistant Professor of Political Science at Arizona State University discussing online analytics. Dr. Hindman uses a variety of data from Hitwise to go very deep into the true analytics and much further than a basic web report.

From the overview, "Using three years of daily Web traffic data, and new models adapted from financial mathematics, this talk examines large-scale variation in Web traffic. These data show that Web traffic is highly heteroskedastic, with smaller sites having orders of magnitude more variation in the relative number of visitors they receive. These consistent patterns allow us to provide reasonable estimates of how likely it is Google will still be the most visited US site a year from now".

How Much Bandwidth Do You Use?

I was doing some research into my various Web hosting accounts the other day to see what services we absolutely need from a host. I noticed that, according to our log files, we are using between 125 and 175 gigabytes of bandwidth each month on one of our accounts and at least 175 gigs/mo. on our other account. We’re averaging approximately 300-500 visits each day.

Granted, on both of our sites, we offer a lot of large downloads, so that accounts for a great deal of our bandwidth. Still, though, I cannot even imagine what we would be paying for that type of usage if we were stuck in an older hosting plan.

How much bandwidth are you using on an average basis? Do the numbers mentioned above sound utrageous, or are they fairly standard? What are your thoughts?

Pages: