Up until a month or so ago I had no access to statistics for this site. I was constructing the site pretty much as I pleased, assuming that visitors would share the same interests and expectations as me.

It was a real eye opener. Some pages were never accessed. Other minor pages seemes overly popular. There was no hiding from what people were really doing while at the site.

The challenge then, is to interpret and respond to the statistics. Change the structure, content and navigation and see what happens! I try to not get too attached to any idea - I put it up, if it works I tinker with it to see if it's performance can be improved. If it doesn't I re-evaluate it. Is it rubbish or should I rework the idea and try again.

I've found it useful to create a 'typical visitor' profile. Where do they enter the site? Which sections do they visit? How long do they stay? and which browser are they using? With this profile created and refined as more data becomes available I can match the site to their needs.

The browser and o/s used by the visitor is going to have a substantial effect on their experience. Apart from the obscure browsers such as linux, my problem visitor is the one using msie 3.*. msie 3.* does not support javascript so getting around this site must be pretty hard! Luckily msie 3.* users make up on average 5% of visitors..

Are your visitors mac or pc based? Do you need to rethink your font and layout choices to optimize them for that platform?

Where will you draw the line? I'd love to include a lot more DHTML in the site but only version 4.* browsers support it. Currently they make up about 75% of site visitors.. It's a bit early to exclude other browsers. Will you make multiple versions to suit different visitors?