On Tue, Apr 29, 2008 at 10:45 AM, Paul Tweedy <[EMAIL PROTECTED]> wrote:
> Yes indeed, and to be open and clear on the purpose of this - the value > in the query string is appended to the item page URI depending the > logical page area in which it appears - Featured, Most Popular, etc - so > we can do clickthrough measurement of how traffic arrives at item pages > and how the site design is performing in relation to the content - which > can inform future iterations/tweaks of the UI to make it better. Plain > old HTTP_REFERER (which we certainly do also have for general user > journey reporting) can't give us this granularity. > > It's a bit of a hack, certainly, but not the worst one we could have > come up with. :) Thanks for the explanation. It's tricky to balance the need/desire to track referals (which may be for good reasons like informing site redesign or more nefarious reasons like tracking clicks from marketing emails). I find the redirect method to be the most intrusive, such as the one used on the BBC homepage, where all the links start with http://www.bbc.co.uk/go/.. Okay, so it's mostly invisible to the browser, but tend to right-click, 'copy link location' and paste into e-mails/blogs a lot - and in these cases it's annoying to then go and copy the 'real' url. The query string method introduces multiple URIs as mentioned above. (I wonder how many blog posts etc link to iPlayer pages with the query string left in). I guess the problem is that there's no 'proper' way to do this. You could track HTTP_REFERER, but that can't always be relied-upon. Or you could use some javascript, but that doesn't work if javascript is turned off. Or you could rely upon cookie-based session tracking, but that doesn't work too well with cookies disabled and is complicated when your site is across numerous servers and sub-domains. Ultimately, I think, you just have to accept that you can't track everything, and go with the less intrusive, more-or-less accurate solutions. With a big enough sample, the data should be good enough anyway. Perhaps the controversal ping attribute is the answer: http://www.whatwg.org/specs/web-apps/current-work/#ping http://weblogs.mozillazine.org/darin/archives/009605.html? Frankie - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
