If you have a power meter, there are a handful of useful metrics to keep track of. First of course is getting an accurate functional threshold (FTP). You can do this in a lab using a step test and blood analysis, or on the street with a time trial. The time trial can be as short as 20 minutes and up to 1 hour. If you do a 20 minute test, multiply that by 95%. 30 minutes by 96-97%, 45 minutes by 97-98%, and so on. 1 hour is the 100% value.
Next - if your meter has a cadence sensor, keep an eye on this. You should be over 80 and ideally approaching 90 on an average ride. A lower cadence recruits more fast twitch muscle fiber (strong-bulk-power type muscle) which burns more glycogen and produces more blood lactate. For a long distance race this is not what you want to happen. It is a bit counter intuitive but using a high cadence / slow twitch muscle is like doing a lot of little quick hops, while riding in a low cadence is like doing a slower series of deep squat jumps. Which one burns more quickly and causes longer lasting fatigue?
You should set up your display so that you can see the percentage of FTP on a 5 or 10 second rolling average (1 second jumps around a bit too much for me), your power, and then whatever else floats your boat. I used elapsed time but you might like watching your MPH.
A critical piece of information can be found by looking at the variability index, or VI, which you can see in Trainingpeaks and probably some other tracking software. This measures how different your normalized power (NP is a wonky way of explaining exactly how hard you really rode. It is usually higher than average power, but could be lower if you coasted down a mountain. See Rick Ashburn's article below for the full explanation.) is from your functional threshold. NP / Avg P = VI.
VI is a key one to watch. For any triathlon type race - disregarding certain competitive strategies like shaking off other riders, or powering past other competitors - you want your bike power output to be fairly to mostly to exactly even. Your effort should be steady. The longer the race, the more important this becomes. The reason is the physiological cost of harder efforts, which demand more from the energy systems and produce more waste. Re: You use up more glycogen and produce more lactate as you approach and exceed the lactate threshold. For long races you want to spare the anaerobic system and energy paths and use fat stores / aerobic metabolism as much as possible.
In a ten+ hour race every bit counts. Joe Friel wrote a few years ago that keeping VI under 1.05 for an Ironman is critical. Various power files on display from professionals and top amateurs usually bear this out, whether or not the course is hilly.
If you do a 5 hour training ride, with the first hour easy, the next 3.5 hours at IM pace, and the last half hour easy, you should be able to select the 3.5 hour section and take a look at that VI. If it's over 1.05, you have work to do on staying smooth. (Looking at the whole thing would include the lower efforts of the warm up and cool down so would be less accurate).
Simply put, when you hit resistance in the form of a hill or wind, rather than push harder to maintain the 80+ cadence, gear up to keep the cadence and power the same. When you have less resistance in the form of a tailwind or a downhill, gear down rather than coast. It's really plain vanilla simple. If you're the type of rider that hammers the hills and coasts the downhill, this will be a tough change for you, especially if you're not particularly aware you're doing it.
Back to NP: This is the main power number you want to go by. If you do a 45 minute time trial (riding as even as possible), let's say your average power was 220 and your NP was 227. Your VI in this case is 1.03. To calculate your FTP you would multiply NP 227*98% which ironically brings it back to 222.46. Use the NP from your time trials to determine FTP.
Onto a couple other notes. Rick Ashburn developed a useful chart explaining the relationship between Ironman bike time, intensity factor (IF - how hard a workout or section of workout is, aka workout NP/current FTP), and TSS (training stress score). I couldn't locate the original graphic online, but Joe Friel explains it well in the article below (with graph). In a nutshell, you want to keep your TSS under 280 in an Ironman, unless you're a super bike pro and/or willing to take chances with the run.
Your IF should be .72 or under unless you're a sub 5:30 rider.
Regarding percentage of FTP target for an Ironman - starting point is 70-72% for 5:30 riders, incrementally higher for faster athletes (75-76% for 5:00) and lower for slower athletes (~65% for 6:30). The longer you're on the bike the lower your effort needs to be because of the amount of time you're putting in. Put this to the test in your training.
Each ride you do get a TSS score. Each long ride you do gets a score. Start to pay attention to all of these. VI and long ride TSS for your long rides, the other numbers just get you to these points.
Joe Friel's article is a good read
Rick Ashburn on power explained
How to analyze a power file from Gloria Lui