I have my Quarq power meter paired both to Zwift as the power source (my Kickr is paired as a controllable trainer only) and to my Garmin, so the data being received by Zwift and the Garmin is identical. However, Zwift reports higher average and norm power and lower coasting time than that recorded by my Garmin when I “just ride”; particularly, when I freewheel (stop pedaling) repeatedly on downhills or following a hard sprint/KOM effort. The the average power and norm power reported by Zwift and my Garmin agree within a couple watts when I do an interval workout where I’m pedaling steadily (no coasting).
The data for my “just ride” session on Friday with some harder efforts (https://www.strava.com/activities/514227124) was as follows: Zwift 181/198/4%; Garmin 172/192/8%, average power, norm power and % of ride time at <10w (not pedaling)
Looking at the respective .fit files, there appears to be some form of response lag, smoothing or over-run of power data recorded by Zwift when I stop pedaling briefly compared to the data recorded by my Garmin Edge 1000. The screenshot below shows a 22 second segment of Friday’s ride consisting of the last 13 seconds of a sprint, a 3-4 second zero power freewheel at the finish and then the start of a bit of soft pedal.
The 2 columns on the far right are the cadence and power data recorded from the Quarq by the Garmin Edge. The 2 right columns on the overlapping image are the cadence and power data recorded from the Quarq by Zwift. Note in particular the difference at between them at timestamp 10:41:29, :30 & :31 where the Garmin recorded 0 power and Zwift repeated the 10:41:28 power and cadence 3 additional times (The garmin also has a bug where it repeats the cadence an additional 2 times when the power is zero). The issue can also be seen at 10:46:54 in the fit file where Zwift again repeats the final power & cadence values 3 additional times before then properly reporting zeros.