Do 191, answer the post workout surveys honestly. Reduce workout targets in workout if it’s too hard. Let adaptive training do its thing. You might be surprised.
I never really tracked before this. I just ride since before this I was doing mostly outdoor rides with mates for fun. With that being said, I think it will be quite hard for me to maintain 190w for 30+ minutes.
I did a ride today based on the “AI FTP” today but it was a recovery ride focusing on cadence so it felt doable. My next hard ride will be on Thursday, so will see how I feel then.
Thanks for the advice. Yeah I’ll try to stick with this for now and see how the next few session will take me. If it’s really too hard then I will do a FTP test to check again.
Not sure about others, but I always dread doing FTP test
my intervals eFTP is 267… Done enough training (with TR but currently not) to know I can do most of the workouts ~300 currently. Only 10% difference, but my point is that intervals.icu is not accurate or at least very conservative.
I have done so many sweet sport workouts, that that’s a good benchmark for me. 4*10@95% for example should be not that hard (for me that is… you will have to find something that works for you)
When you are new at training, just start with a ramptest. Might also be not that accurate, but AI will fix that for you by starting with low powerlevels. It won’t be overestimating more then 5%
Intervals.icu uses a model curve to predict your eFTP based on your best ‘max effort’ from the data you have fed it (you can set the time period this effort should be based on - I think the minimum is 5mins, but most people would argue to have it set for longer). This means its estimate will naturally be low unless you have recently gone truly full beans for a single effort of whatever time period you have set. Based on what you say about ‘riding around with mates for fun’ that may well be the case here, I don’t know.
It also means it will be lower than TR AIFTP based off the same data. The workout Lamarck is often cited as a good test to see if your TR FTP is ‘right’ - but Intervals will only see one 10 minute max effort, not four, and fit to a curve accordingly. And, 4x10 @100% does not necessarily translate to 1x30+ @100%, so many will argue your TR AIFTP is ‘too high’, but let’s not get into that here.
TL:DR - unless you’ve gone full gas at some point recently, Intervals has you too low. If 191 intuitively feels too much, TR AIFTP has you too high (for a 30minute effort at least). Either do a ramp test, or pick a number in between the two figures for 10 workouts and let AI do it’s thing.
Hey! This is a great question. I think this difference mostly comes down to how different platforms use different methodologies to estimate/ detect FTP.
As @Flashpoint51 mentioned, Intervals.icu may rely on peak maximal efforts, which leads to a more conservative estimate—especially if you haven’t recently gone all-out. This difference in methodology can sometimes explain variations between AI FTP Detection and other FTP estimates.
TrainerRoad analyzes your training history, biometrics, and performance to provide a personalized and accurate assessment of your Functional Threshold Power (FTP) by analyzing your training history, biometrics, and performance data. Unlike traditional single-effort tests or estimates, AI FTP Detection offers a comprehensive and continuous understanding of your current fitness level.
As @jamesdrichey suggested, I recommend starting with your TrainerRoad AI FTP Detection result and ensure to provide feedback in the post-workout surveys. This helps TrainerRoad ensure that your workouts are always appropriately challenging and aligned with your fitness level. TrainerRoad continuously responds to your evolving fitness and the data we recieve about your training and performance.
If anything feels too hard or too easy, your plan will adapt accordingly.
Give this a shot and let us know how you get on in a week or so .