I’m thinking about trying out TrainerRoad, but I have a question regarding the difference between indoor and outdoor training.
I do nearly all of my structured training indoors, try to get my Zone 2 done outside (can’t bring myself to ride more than ~90min indoors) and do also the occasional group rides outside. My question relates to the differences in power and HR between indoor and outdoor rides: I can generally produce higher watt numbers outdoors than indoors. The difference is fairly significant, even though my indoor setup is set up with an AC and I use also multiple fans. I heard once that the reason might be the trainer I use (Tacx Neo 2), but whatever the reason is, it is real.
For my training it doesn’t make a big difference, I go to the limit for the interval duration in the workouts that I’m running and I’ve seen very good progression over the years. The logic I operated under was always that in the end my muscles don’t “know” what number of watts a display is showing me, but that I’m sending a strong impulse if I ensure that I do the intervals at the maximum effort I can sustain for the prescribed time.
My “problem” now is that I’m not sure how the TrainerRoad AI takes differences between in- and outdoor efforts into consideration. If it bases a lot of the workout numbers on the power I produce on my outdoor rides (on a climb for example), it will result in my structured workouts being too hard. The flipside could be that if it bases everything off my indoor numbers it will read my group rides and outdoor efforts as much more fatiguing than they were and have trouble judging how much rest I need.
At the end of the day I’d like to know if the AI is able to track indoor/outdoor efforts separately and understands the difference or if this is something that will screw my prescribed workouts up for me.
This is a common question we receive, as many athletes perceive a difference in their ability to produce power indoors vs. out. This isn’t always consistent in one direction; speaking from experience, I often feel stronger outside when I haven’t ridden on the trainer much. But after a long winter indoors or after training specific skills on the trainer, I then perceive the opposite and initially produce what seem like lower numbers outside for a bit.
That said, TrainerRoad AI does a really good job of handling this. The AI learns from your performance and understands your strengths and weaknesses. It absolutely takes all the riding you do into account when understanding your fitness, but your recent structured work in a particular zone is a major factor in guiding your next workout in that zone. If you do an indoor VO2 max workout and then go ride outside the next day, your performance in that VO2 max workout is still going to be a very important part of how the AI considers your next VO2 max workout. The outside riding is absolutely a factor too, but it’s understood in context. And if, for instance, your outside riding caused the AI to recommend an FTP increase, it does a fantastic job of calibrating your workouts at that FTP to be harder or easier, as needed.
To again speak from experience, last year, after building up my fitness indoors in the spring, I rode almost exclusively outdoors and unstructured through a good portion of the early summer. With that fitness I’d built through my structured workouts I was regularly setting power records on the local group ride and in races. In the past, this would often mean it’d take a few workouts to get my indoor intensity right when I started back up with some indoor structure. This year, I started testing TrainerRoad AI mid to late summer by adding in some indoor workouts, and it was uncanny how well the AI dialed me in. My fitness saw a huge bump from maintaining this structure and I never had a workout that felt too hard or too easy.
I think you’ll find the two can work together quite nicely, even if your abilities seem to differ one way or another. The AI is extremely adaptable and responsive, and this is one of the use cases it’s proven most effective at handling!
Really interesting. Do you think it’s possible that this is a case where some of what the AI is doing fall into the ‘black box’ category, where it’s recognizing signals/patterns independent of being explicitly instructed to look for them?
I wouldn’t quite characterize it that way. TrainerRoad AI is designed to identify each athlete’s abilities on a more granular level than we’ve been able to in the past. So I think in that regard, it’s doing exactly what we want it to.
Sadly TR doesn’t offer a way to isolation indoor and outdoor rides. An often requested feature is a way to do this in the season compare power results (E.G. show me my power curve for indoor rides for 2025 vs outdoor rides) in order to see the difference in results. It has never been implemented and TR assumes that you’ll manage to make the indoor and outdoor power numbers close enough for it to not matter too much. I’m like you, I can maintain better numbers outside than inside (better bike, better cooling, perhaps the indoor resistance isnt perfect) so I always feel like if I do lots of indoor riding TR is seeing my indoor numbers and over reacting, and vice versa. There are other programs that make it very easy to quickly compare your indoor vs outdoor power curves for different periods, so I now use those to keep a perspective on “where I’m at” capability wise in each environment.
This is a question I raised during my Onboarding Call as my indoor and outdoor power curves are significantly different (using the same power meter - pedals).
So far in 2026 I have been exclusively riding indoors so it has not been an issue - but I hope to get outside soon, so we will see what happens then.
Intervals.icu is great for this - see the pdc below for my 2025 season…
I can relate with the remarks mentioned above. Me too am experiencing a difference between the outdoor and indoor power numbers I’m able to register. My indoor setup is different than my outdoor setup so I don’t know whether this difference is due to different power meter sensors, calibration status, mental status or just the weather conditions.
In the TrainerRoad application the indoor/outdoor difference shows in the TrainerRoad AI FTP detection. The detection is over-estimating my FTP by +/-20W. Already with the current FTP setting my indoor workouts are very hard and barely possible to complete so unfortunately I can not accept the FTP detection. By consequence I’m also not able to use the AI FTP prediction feature.
I can also clearly see in the new “Power Records” feature that the best efforts are always linked to outdoor rides, strengthening my assumptions.
It would be nice if TrainerRoad would add an “FTP calibration offset” per workout type to mitigate this problem.
OT you got me curious about my indoor & outdoor. I seem to get far more power outside for short durations (whether the power meter can be trusted for that or its just spikes ) but over longer durations I can get slightly more out on the turbore I think because I can focus more and I’m not racing outside anymore.
Many of my records are inside. 5 minute power record is almost always in a Ramp test. Im not sure how it would be much more otherwise. Your curves should converge there.
Same with zwift races since there is no real drafting. I have races that normalize far over my FTP. Zwift Crits are often about as how as 20 minute power gets.
) but over longer durations I can get slightly more out on the turbore I think because I can focus more and I’m not racing outside anymore.
