I like that deep pages are being counted - many people were arguing for front page only - but we’ve written lots of stuff that is well linked and visited often (like this - even Blogger linked this one!) and they are valid visitors in my book.

I also like the fact there is a margin for repeat visits during the day (without the opportunity to abuse it) as we update quite a bit (as you’ve seen the last couple of days! so repeat visits are a must for New Links readers.

I think I’ve got a neat solution for processing the data offsite now too. So we’re getting there. Hope to have something usable in the next couple of days now…

(ohh and hi! hope you don’t mind me crashing on in here!)

I’ve had some chats with folk this afternoon, and this is what I’m currently thinking:

People put the following on their pages:

<a href="http://www.britblog.com/">
<img src="http://www.britblog.com/images/icons/some-icon.gif?uid=123" />
</a>

The uid parameter is the member’s ID, and the icon is actually a script that returns the image with cache-unfriendly headers.

So far, so good.

When this image is requested, the client IP address will be logged in the server access logs, along with the referring URL (the blog page that is being viewed), and the date/time of the event.

I can process this file nightly/weekly and pull out all these lines and do the counting. Multiple entires for an ‘IP address/blog page URL’ combination within one hour will only be considered as one impression. This way multiple requests from the same reader on the same page within one hour can be discounted. It also makes it harder for people to trick the counting. The most anyone can easily spoof will be 24 impressions a day, which will be quite insignificant.

Any comments?

Our BritBlog button is part of our template and is on every page anyway.

To do a correct analysis of readership, the icon should be present on every page (post pages, home page, category pages etc.). But hey, it’s all good publicity for BritBlog