Does anyone elses Garmin Connect calculate average power differently to all other analytical services (Strava, TR,, etc)

Hoping someone can can offer some explanation to this…

It’s only become apparent to me this year and it’s only really obvious in races, but Garmin Connect seems to calculate average power differently to everyone else, despite it all coming from the same source (fit file?)

So today’s race was recorded by my Edge 530 and initially uploaded to Garmin Connect:
Garmin Connect = 236w average power
Strava = 225w
Trainerroad = 225w = 225w

I could handle 1 or 2 watts, which is generally the case in other rides, but for races there are these large differences.

Anyone know why or notice the same on their rides/races data?

Surely average power is a standard formula with no variables such as FTP, body weight etc.

Is Garmin set to include zero’s in its average?

yes it is

Are you having ant+ dropouts? Strava counts dropouts as zeros, while some other sites don’t count the dropouts against the average. I can’t remember off the top of my head which way Garmin handles it.

Ahh this could be a good point!! Is there an easy way to identify this?.

There were a few drop outs in HR data but power was there during these periods.

Have a look at your times. You’ll notice they’re all over the place depending on the app.
Could that be the cause?

Strava v Garmin. Garmin is correct.

Looks OK barring a single second in the garmin :man_shrugging:t3:

Although this has baffled me in the past why the times can be significantly different

1 Like

And I thought I had a dream run with only one traffic light stop. :grinning:

Stuffed if I know what’s going on. I’ve always looked at the different ride times listed and figured things must be out if they can’t get that bit right.

:grin: it was a road race so in theory it shouldn’t even be one second out.

But i think drop outs is the most logical explanation and i can only assume that Strava/Trainerroad/ etc all count these these as zero whereas Garmin look to exclude them or maybe fill with a 30 sec rolling average. So is Garmin actually the more accurate, despite being in the minority?? I’m just struggling to find anything online to support this.

As mentioned, it’s only obvious in races, so i’m thinking with 60+ other riders also using ant+/bluetooth connections is why there is the increase in drop outs…

Just wondered if others had experienced this??

1 Like

I looked at my last race, and have the same. Garmin Connect has AVP 6 watts higher than Strava and other tools. Ride time is the same on all platforms. Weird. (Using Edge 1030, Stages LR power meter).

Edit: total work (kJ) has exactly the same discrepancy between Garmin Connect and Strava. So maybe GC is indeed backfilling dropped records or something like that.

Sorry mate. That’s what I get for skim reading and multitasking.

I checked some old race data and there’s only a few watts difference. Nothing as bad as yours.

Yes, I also have the same discrepancy with ‘Total Work (kj)’ as well as NP. Would be nice to know what method Garmin are using and determine which is the more accurate.

According to this thread on the garmin forums: : The power meter sends accumulated work (kJ) data in addition to the power readings (Watt) for each record it sends. So in case the headunit misses some records, it can use the accumulated work to still calculate average power correctly. Apparently the headunit and garmin connect use this data for that purpose, while strava, trainerroad, and many other tools do not. In summary: the data Garmin gives on the headunit and in garmin connect is more accurate.

I may try to switch to bluetooth (now using ANT+), because I think in theory bluetooth should be less susceptible to dropouts.


I agree. It all has to do with dropouts and Garmin does a really good job at filling out all the possible data fields of a fit file, while other companies dont. This is part of the reason why garmin fit files are bigger. Garmin fills in alot of summary entries (activity, session, lap). So in GC they just read these headers to display your summary data. Because other sites like strava have to process fit files from many headunits, they have to process the actual 1sec data entries to compute the summary data instead of using the summary entries because some don’t put data in them. I actually prefer the strava method as it’s a good indicator that dropouts are occurring.

1 Like