For me it’s also about 5-10% (on a clean bike). I did NOT test discrepancies inside, but I have also assumed that outside it is just easier to deliver more power because you are propelling a device into motion, whereas indoor you are completely static besides the momentum of the flywheel.
So endurance indoors is 155 watts, whereas outside it is 165-170, for the same heartrate given the same freshness (endurance is where I spend most my time outdoors, so seems best comparison for me).
Different trainer for me (a Suito); I never looked closely at the data but the 4iii and trainer were close at low to high cadences but at very high cadences the trainer was well below the 4iii. It was the same with my Favero Be Pro S and the Suito. Makes me think it was the trainer that was laggy 
This only work if your discrepancy is all time the same. Most of time you see in lower power burger gaps than around FTP, zo that workaround doesnt work.
The best solution is sell your 4iii and bought the same powermeter on your other bike.
Tail wind? Different air densities ? They might balance the maths out?
That’s a good point, and it did seem closer %-wise at FTP and higher powers. I’m not super concerned with 100% accuracy as I just want to have consistent power numbers to compare to- that said, I’d like to minimize the gap for intervals so that my wattage targets, TSS scores, and so on will be comparable to the quarq I use to set zones and calibrate training.
Of note, my quarq does typically show a 53-47 right-left balance (slightly more at higher powers, less at low powers), so adjusting the multiplication does seem at least partially justified.
Answering for myself only - It takes a lot of time/setup/prep/etc to get things lined up. Clean chain, cassette, spindowns, correct installation of meters, bedding in, zero-offsets, static weight tests/verifications, in-sync data recording using head units, making sure there’s no dropouts, having a test protocol that tests a wide range of use-cases, using a straight chainline for steady-state ERG, ensuring I’m riding balanced (not wonky AF with a 40/60 average balance), only comparing apples to apples (ie, LEFT only with a single sided pedal), digging into the data at different levels/averages/etc.
tldr; It’s a total pain to get everything lined up.
If things don’t line up within the specified % spec, I usually go back to the manufacturer first (and second, and third… or fifty times if you’re Sigeyi). Depending on their response I’ll either do a video/review or simply send the power meter or trainer back and move on. I’m ok with a ‘few watts’ lost in the drivetrain… but nowhere near as much as some people are happy to write-off as being acceptable.
Has anyone else had this problem? Scale factor doesn’t seem to stick. 4iiii said they are working on a fix.
I have Suito and s-works dual power cranks (which is 4iiii actually) and also have discrepancy but is not constant. At 150w there is no different, and on 200w is a 10w, at 250 is 15w, at 300 is 30w…
(Suito reads lower)
Like I was saying above. I think the Suito lags behind at higher cadences. It was similar to my Be Pro S (the Suito is playing catch me up for higher cadence/ power efforts).
This info is very intriguing to me! There probably isn’t anybody in the world that looks at the minutia of bike power and drive trains with a more calibrated eye than you, GPLama! I mean, there may be more precise measurement setups but I don’t think there is anybody else who has as big a dataset as you do…maybe ray.
So it’s interesting that you are able to observe only a few watts drive train losses! Let me just say: you could really be on to something here if we can figure out why that’s true. Do you have any thoughts on the matter that you’re willing to ‘give away’? 
I know when Jason published the 1x/2x loss data* I quickly followed up because THAT dataset was at odds with data I had previously purchased from him**. The old data was 5W to 6W more efficient than the new data. Were new drive trains actually that much worse? Ha! No…it turns out if you read the older reports & the velonews article carefully enough, the velonews data was taken with a derailleur in the system. The old data was taken without a derailleur in the system.
I thought, ‘Just the derailleur is 5W?’ (look at the 53-11 annotation on the appended charts). Even @53-15 the derailleur is sapping a few watts…a little over 3W.
Any thoughts on why your drive train setup is more efficient than Jason’s? If you can figure it out you might just have a tidy little business on your hands.
**
A front derailleur that isn’t shifting isn’t causing any drag as it’s not touching the chain
Is it just me or are some of these complaints not really valid? When you compare a left only power meter to something that measures power from both legs you can’t really be sure what your left right balance is
I’m not sure people don’t have a valid issue, but as you’ve raised there certainly are reasons for the differences.
That said…if a left side only PM isn’t accurately reflecting power is it really an accurate power meter?
It is, as long as you are a machine that is 50/50% power at all power and fatigue levels 
The fact that any single-sided PM doubles power from one leg is a known issue, but far to often ignored in these types of power difference questions.
It’s clearly not as accurate as other PM’s that take true full or left-right power, that is certain… if you assume we are not perfect machines (a safe assumption IMO).
I have a Kickr and LH 4iiii set up which I’ve used for a few years. Kickr read higher to the point where I stopped using the 4iiii to control the workout. Few months back ran into a bunch of ANT+ connectivity issues and as part of which did a full factory reset on the Kickr which saw the Kickr give me lower readings (my apparent FTP got trashed!) and now means the Kickr and 4iiii are much closer so I’ve reverted to using it to control the workouts.
Still trying to get over the blow to my FTP though!
You guys are scaring me, my 4iiii factory install crank arm is on its way back to me from Canada. But I’m just looking for a little help knowing what’s going on for my outside rides since HR isn’t enough data for me. I’m not sticky on the absolute values, just as lost as I get some precision and interesting data that will help me increase fitness and pace long rides.
The 4iii is working fine for me and I just ignore the turbore. I think thats the mistake a lot of folk are making, wanting everything to read power identically but there’s a lot of room for +/- s between devices. Just pick one and train by it. The 4iii for me is a good one 
Thanks for the reassurance. I plan to get some Favero Assioma pedals down the road to use on my road and TT bikes (TT bike lives on the trainer), so I know the values between 4iiis, the Elite trainer, and the Favero pedals will vary. Anything is better than having no power data from outside.
What I see/record is all I have to work with. As for claims/publications that are usually backed by a company selling low friction things or magic dust… shrug… I’d love to independently report that I see the same losses/savings/etc as they report… and I’m sure these companies would love that… but I just don’t see the same numbers.
