Subdirectories v/s Subdomains in URL structure

Deciding on a URL structure for your web properties is perhaps one of the most important things marketing has to do. The same problem also applies to web products which are not behind paid walls. To better know the difference between these two approaches for determining which one to pick, I'd refer you to Matt Cutts blog. Now to go on to why you should pick either one of them...

The right answer depends on how mature (or old) your website is - how much traffic you are already getting and whether the benefits of subdomains apply to your situation.

The safe choice is always using the subdirectory model. This works well for SEO and managing the website. The main reason why you may want to go to the subdomain model is that with some good keywords in the subdomain, it is slightly better for SEO. From a branding standpoint, it puts a stake in the ground as well. In other words, it sends a stronger message. Beware - there are risks involved and the good news is that you can actually do a slow migration if you decide to head in that direction.

Search engines treat subdomains as separate websites. It just doesnt finish there. They do special handling for sub-domains to detect duplicate content. Because they know that all the subdomains are usually owned by the same entity, they want to protect users from spam. So while posting content, you must be careful not to unknowingly duplicate stuff. Its always better to keep the content split in different disk areas or web servers to avoid this. Any SEO hit due to an accidental sharing of data can have a long lasting effect.

If you do want to experiment with subdomains, one method would be to start at the least risky component of the website. Note that subdomains also demand that there be substantial content for Google to see in that domain. When you split the domain off, you will loose traffic which you now need to build back two times for the ROI. On the technical side, if you own the complete server infrastructure you may want to play around with a reverse http proxy using apache or lighthttd or nginx etc which allow URLs to be mapped w/o physical changes in the network.