So you have a power meter and have begun collecting data--now, you wonder what you are supposed to do with it. In a lot of ways, power data is not the same as data collected by a cycling speedometer, or even a heart rate monitor; while you can track power data as you race (many riders will do so to help with their pacing), the real value of power data is tracking changes over time. A power meter will help you determine whether you are getting stronger and fitter as a result of your training. It also will help you optimize your aerodynamics, though this is a bit more complicated.
The very first step for anyone interested in power-based cycling training is to purchase a copy of Hunter and Coggan'sTraining with Power (I tend to discount their views on HR/PE) and pay close attention on what normalized power is all about. Normalize power is related to, but not the same, as average power. For instance, in events that have little variability like coasting and drafting (such as a time trial), then normalize power and average power are pretty much the same. In events in which there is a good amount of variability (such as in a criterium), it is not unusual to see huge, but short in duration, spikes in power as riders jump out of corners or sprint for primes. Apart from these bursts, a clever rider might spend much of the race pedaling lightly in the draft of other riders (I know about this!). While the average power of a sprinter in such races might be relatively low, it does not accurately reflect the actual load placed on the rider during the course of the event.
Normalized power attempts to equate all efforts as a comparison to a one hour, full time trial effort. There are lots of algorithms that enable a ride to calculate actual work load, the easiest method is to purchase a copy of Cycling Peaks software that automatically calculates normalized power. The normalized power value, then, can be used to compare efforts of very different types of events, independent of the variability of average power.
A power meter really is an investment in time. What I generally recommend is that you track normalized power for lots of training sessions, paying special attention to HR/PE and how it correlates to the normalized power data. Once you get a sense of your normalized power for events of different durations, you can use these values to help you pace during the course of a specific race. For instance, you will use your power meter to track efforts on the climb and monitor output on the down hills, so you don't go too hard downhill, or too easy up. On top of all of this, you'll be paying very careful attention to your HR/PE as your final determiner of effort.
What's interesting is how you can use a power meter to manage your efforts on courses of different terrains. For instance, if the course is fairly flat, then you can ride at your normalized power value (which in this case will be similar average power) and be assured that that you should finish the big leg relatively strong; if the course is really hilly, then you would use your power data in a very different way. Here why: when riding downhill, the biggest factor that you face is aerodynamic drag. Drag increases as a cube of speed; what this means is that pushing 30-40 more watts going downhill will get you only a slight advantage over your competition, because drag at speed is so high. If you are climbing, a 30-40 watt increase makes significant gain over your competition, as aero drag is minimal.
Example: If you're doing an Ironman on a hilly course and you calculate that your normalized power for this distance is approximately 175 watts, consider going a little harder when riding uphill, and a little easier down hills. Because everyone pretty much goes fast downhill, use the uphills (within reason) to put time on your competitors, while recovering on the downhills.
Power-based training can be incredibly complex, but the time invested in tracking data over time can yield significant results. Training by power is not necessarily for everyone, and it is possible to get extremely positive results by tracking heart rate, too--in fact, I still tend to do this for most of my workouts, using my power meter for testing and to track long-term changes in performance. For more information on power-based training, check out Hunter and Coggan's Power 411: http://home.trainingpeaks.com/power411.aspx