As with Max HR, you test for your FTP. You can either suck it up and suffer the full one hour maximal effort test (on the assumption that the definition of FTP is your maximum 60 minute power, which is not the only definition) or you can approximate it by using a shorter test.
For the most widely used short test, you will essentially do a 5 minute maximal effort, then recover, then proceed with a 20 minute maximal effort. You take the average power for the 20 minute effort and subtract 5% to ESTIMATE your FTP. The exact test protocol is detailed in the book 'Training and Racing with a power meter' by Hunter Allen and Andrew Coggan. Which I would suggest you buy and read.
The per kg value people refer to is simply power divided by your body weight in kg.
Don't get hung up too much on one value, and certainly not to 2 decimal places! For most of us mortals, the main benefit comes from doing regular FTP tests to determine how your fitness is improving (or falling). It's like your weight; a longer term rise or fall is more important than day-to-day variations.
With this in mind, it's probably more important that your bkool is consistent (same effort gives the same power value) rather than accurate in absolute terms (200W from your legs shows as 200W on the screen). The latter case applies if you're racing against other riders on bsim, but that's another story!
Any training device measuring power is going to be subject to variation due to setup, calibration, temperature etc. On top of that, riding a 'good' FTP test takes a bit of practice. Even then, you may or may be a bit 'off' one test day, or you might be better in the morning as opposed to the evening. The test protocol will also give different results; I tend to score a bit better on an 8 minute test than the 20 minute test mentioned by @Rob3rt. Getting your hardware setup and technique as consistent as possible will give the most useful results. For roller-based resistance unit devices like the bkool, try and keep the tyre pressure and force applied against the wheel by the resistance unit the same for each test. For this reason, getting a consistent setup is always going to be more difficult than 'wheel-less' devices like the kickr or wattbike.
As @Rob3rt pointed out, the bsim test protocol isn't that useful as, from memory, you're straight into the 20min effort. Unfortunately, as the bkool is tied to the bsim software, it's the only one you've got. TrainerRoad has 8min and 20min tests, and I'm sure there'll be other versions on other training-focused software like training-peaks, if you want to see a more representative test profile.
I'll maybe do an FTP test approx once a month and have done for about 3 years, first using a turbo and a powertap, then more recently a kickr. Although I've upped my FTP significantly in that time, typical test to test variations are around 5-10W on the kickr and 10+W on the turbo. My most recent tests on the kickr sit around 290W (about 3.7W/kg), but I've had one result over 300W. I always do a spindown recalibration before every test.