I’m figuring this uses the so-called “Chung method” where you calculate differences in drag based on power outputs for repeated out-and-back time trial. But this does the math for you which is nice, even though you could just as easily do it yourself in Excel/Google Sheets. I’ve not tried it but I like it - as I understand it your power output doesn’t need to be precisely the same for each effort, it just needs to be relatively close. It does require either a completely windless/traffic free day, or at least a fairly constant wind speed and direction. Otherwise that will contaminate the estimates.
If you like math, here’s a blog post where a rider tests the accuracy by fixing different sized styrofoam spheres to his bike, and then he calculates their size based on the changes in predicted drag coefficients for each run. The upshot is it does a pretty good job.