Pre ballot browser stats


These are the browser stats for a popular Dutch travel blog where I work right now. The special thing about these stats is that they were derived from a test that tried to get data on the usage of Javascript with our users.

Our audience is Dutch, absolutely non-tech and our user base is more girl than boy. For these stats, the logged in users were not counted, because they are abroad in all kinds of hell (read: internet cafes), which usually have crappy computers with crappy browsers, depending on where your vacation is at.

So this audience is basically the friends, parents, grandparents and other family of very non-tech users. I believe this is a very interested target for a lot of websites. So these stats might be of some interest to you :)

The test ran for six days, Tuesday morning till Monday morning. In total there were 1.75M hits that count for this these stats. Divided as follows:

Test ran: last week of February, 2010
Total hits: 1786892
79.4% IE (v8: 49.4%, v7: 24%, v6: 6%)
11.1% Fx (v3.6: 2.5%, v3.5: 8.5%)
4.8% Chrome
4.3% Safari
0.3% Opera

It was quite shocking to see the Opera numbers actually. Some of those hits are probably my own :/

As this test was run to test for Javascript support, I'll explain how it was set up. For the record, the Javascript support is nearly 100%, which I think is good news.

Every webpage on the website received three content elements to be loaded for this test. One regular image, nothing fancy. One image embedded in a noscript tag, this one should only be loaded for clients that have Javascript disabled, or don't support it at all. And of course one script tag, loading an external script, testing Javascript support.

Each of those resources led to the same PHP script. Upon the page request self a record would be added to a table in the database. The record ID would be used in the url to fetch the additional resources and to recognize the request.

When either of the three resources was requested, a counter for that resource for that record was incremented in the database. Plain and simple. In the end, for virtually all valid regular users, there should be either one image and one script or one image and one noscript image downloaded. Bots either completely ignore the resources or they download the image(s) when they are looking explicitly for them. Apparently the noscript images are ignored in that case, since a single request to both of them would have inflated the amount of "users" that don't support Javascript.

And for this test, only those records were counted who requested only two resources, one image and one script. The hundred or so valid noscript requests are ignored in this case. Deal with it :)

So, for NL, these pre-ballot stats clearly show IE is still on the upper hand... by far. I'll run this test again in two weeks. Let's see whether any significant changes will have occurred... :)