but I don't know what area's in pen testing are important / more important than others; or if it's purely a preference thing. It seems you can go into forensics, network testing, wireless testing, web application testing, exploit writing etc.
I'll break each out separately based on how much time I spend on each. Keep in mind though, this is a field where you should really try to learn everything you can. However, hopefully this helps you prioritize...
Forensics: 0%, we have a dedicated forensics guy, that is all he does. I think this is normal, but others can correct me if I'm wrong.
Network testing: TONS of TIME, a typical pen test in my company is some flavor of a network pen test, i.e. external, internal
Application Testing: TONS of TIME, I spend equally as much time on web apps as I do on network PT. In fact, on almost EVERY network pen test, you'll run across web apps that may/may not help you with the objectives of the pen test. You need to know both network and web apps equally.
Wireless: SOME, these are usually "bundled" into internal pentests our company sells, but it's not the norm. Not near as much time is spent here. Wireless is pretty easy stuff though, it really doesn't take all that much time to get up to speed on the basics, as well as enterprise wireless stuff.
Exploit writing: (as a hobby, LOTS) (for work, much LESS) If a company has a true research group, you might be doing this all the time. Similarly if you worked at a place like Metasploit, you would be writing sploits all the time. We have a lot of guys at my place that have never written an exploit, its not a "critical" skill for basic pen testing.
In my opinion, I think you should start with network and web apps, study them equally, become a bad ass at both. The rest will come easy if you master those two.