75% of the decisions into layout are due to web trends and three years of analyzing the site. The other 25% is what I think looks 'good' ;) I currently use SiteMeter, Google Analytics AND Woopra, but I'm probably going to drop SiteMeter for lack of useful data.
The logo being big is because recognition degrades the smaller a face is, mostly. 150px is about as small as it 'works'. I used to have a 100px image and people kept asking what the image was. For a 'professional' site, I'd do something small, but as a fansite, 'big' headers are 'better' - according to polls and trends. I bow to traffic results.
The lack of categories was born from organization. Basically you get 4 cats: Jorja Fox, Site News, Community Stuff and Featured. As far as searching on that, having the category was pretty useless for the every day viewer. Now, the tags got a lot more traffic so that was easy to decide. Obviously it's gonna be different for everyone! :)
I'm still experimenting with ad placement. The 728wides as a banner ad get more bank for your buck, but they do mean the browser has to be 'wider.' I hand smacked IE 5 and 6 to go with the smaller ads and a fixed width, since they basically suck, but a stupidly high number of my visitors use it (60% of visitors use IE and of that 23% of the 60% use IE 6 and .2% use IE 5). So really not a lot at 14% of all users, but enough that I don't want to tell them to FOAD. The site does look best in FF or IE 7/8 or Safari or pretty much any 'modern' browser. Thanks to BroswerShots, I was able to see how it'd look on different OSs. It is, honestly, the most work I've done on a website for design since ... 2005? It's the most complicated one.
But honestly it was faster to mock this up in Hybrid News than it was to mock it up in Structure (which I'd been doing before news came out). Obviously I customized the heck out of this. I changed fonts, moved layout and such, so really you can tweak this to your own needs :)
Now I'm agonizing over widgets vs manual changes. Probably go to wigets.