One of the main themes both of this site and of many of the discussions that take place in the comp.info.www.authoring.* Usenet groups is that of structuring web pages properly.
It doesn't however take much searching around the web to discover that many webmasters are not convinced of the benefits of doing things properly. And quite a few contributors to c.i.w.a.* take the attitude of “it works, so why should I worry?”
A few years ago I had the opportunity to put these structural ideas into practice on an existing commercial site (the site has recently been completely redesigned, so there is no particular point in linking to it).
When I first met it, this was a Javascript-dependent, frames-based site with poor HTML markup (for example superfluous layout tables and headers marked as paragraphs). I got permission to restructure the pages, as long as the basic layout and colour scheme remained unchanged.
I created a new HTML/CSS template, which made no use of frames and initially no use of Javascript. The text was copied page-by-page from the old site, and the interspersed markup tidied up. I subsequently added a CSS/Javascript drop-down menu at the request of the site owner, but Javascript was not essential to navigate the new version of the site: it just gave the option of skipping one level of index pages.
Apart from applying basic good markup principles, the only optimisation I applied was to convert the existing GIF images to PNGs, which saved around 10% to 40% of the size of each image.
The benefits realised were:
Those are the benefits that were acknowledged as such by the site owner and his marketing managers. In addition I could mention that the site looked distinctly better than the old one in browsers other than Internet Explorer.
All in all, it seems to me sufficient reason for doing things properly.