The real problem is in educating the client as to what is most important and what will deliver the best outcome. This has saved us a lot of time since we’ve adopted this approach on recent projects.

Squad has been working towards flushing out the content and Information Architecture early on in some simple bare bones XHTML pages. This helps explain what content is expected to be on each page and helps organize the site. The upside to this approach is that you’re already building something you’ll use and can touch and feel.

We then move on to some basic layouts of the site by adding in some CSS to the bare bones pages. Changing CSS is much more efficient than working in Photoshop and really helps you really nail the interface (if we’re talking XHTML sites) and expose usability problems.

At that point we go into photoshop to create a design that looks great off a foundation that we know works and flows well. From there integration of the design is much quicker since you already have a foundation built and exposed many design issues. This makes sure you’re building something you’ll use each step of the way and you’re not creating something that can’t be done efficiently.

This helps the client focus on what is important & see it come together , exposes many of the different actual states by using real content, lets you quickly adjust elements for comparison and lets you start building before the design begins as the design should support a solid foundation.

This is somewhat different if you’re talking Flash based site but I’m guessing we’re talking HTML sites here.

Thats my 2 cents.