Yea, it is a huge change. However, I dont want you to be alarmed by it, because I’m using Piwik in an unconventional way (as a third party embed).
I didnt add anything to the source code. I’m using a convoluted, hacky way of resolving this.
see 301 Moved Permanently
Anyways, the hacky way involves putting the piwik analytics code into a file(whatever.js), setting a mod_rewrite rule to always serve the file whenever the url matches something like http:// myserver/trackit/…, then including javascript in the openx adtag to write a 1x1 iframe with said url, and appending url params containing top.document.referrer and top.location.href, and modifying the piwik analytics code in the file to use/honor these params using setCustomUrl() etc. Also setting a P3P header similar to what Openx uses (its a standard compact string like CUR ADM OUR NOR STA NID)
If you read the previous paragraph, you’ll agree that this is a hack, and not an elegant one. It creates hurdles at the same time as it solves the above.
Before this hack, what I would have done is to add the piwik analytics code into the append/prepend area of a banner/zone on openx. No iframes involved. The ad itself was served via openx js tags(again no iframes), and I was missing all the IE visitors.
So again, unconventional case (which might start becoming conventional as ppl clamour to get ad stats beyond what an ad server provides).
If you’ve noticed recently at adTech, Google Analytics had a bigger presence than doubleclick/Adsense. They’re really pushing for analytics in advertising and there are tons of people hungry for that sort of data, i.e. where was my ad served, what article, what demographic of ppl viewed it, etc)
I’ve been using a similar hack for a while with google analytics, but I think they dont allow things like setCustomUrl(), so it muddies the data collected.
re demo.piwik.org “but maybe these are not all visits?”
Perhaps the % of IE visitors on demo.piwik are verifiable by another authority (or access logs)?
For the site I had the hack on, the number of IE visitors were more than FF/Chrome/safari (as seen on the second attachment) and that seems to be a more realistic percentage for my site.
Also, the total number of “browser types” was 64 (which is quite high compared to the demo.piwik as well). I’m assuming these arent bots, as the bot tracker plugin only shows 3 bots regularly visiting my site.