Advice on 20 Min FTP test result v.s. AI Detection

Fellow riders, I just did my first 20 minutes FTP test earlier today. Ever since I started using TR (since 2 years ago), I have been doing only ramp tests and started using the AI detection since the beta testing days.

In the past 4 weeks, I’ve been able to finish 100% of my workouts (mainly SS with a day of VO2 Intervals every Tuesday). I’m on a medium volume plan (8-10 hrs a week).

My test result came out to be 200, which is lower than my before-test FTP of 206. I got curious and ran the AI FTP Detection and it predicted a score of 209. That’s quite a big difference. Unlike my ramp test results in the past, which typically just differ by 1 - 2 points.

I realize that I’m new to the 20 min test and might not have done a proper pacing. I’m interested in hearing from you guys who have more experience with doing the 20 mins test, and had a chance to compare the AI detection accuracy to the test results.


It’s not a huge difference. You’ll have a normal day to day variation in threshold based on fatigue/freshness.


Pacing takes practice.


The difference is not that big, 4.5 %.

Most likely, you just have to get better at pacing specifically for the 20-minute test. That takes practice, unless you are regularly doing 20ish-minute efforts. Just that alone could explain the discrepancy.

You wrote that you have been using AI FTP and you have been able to complete all workouts. Then I suggest you just stick to AI FTP’s suggestion and continue your workouts. If you are able to finish all workouts as prescribed, just stick to AI FTP’s suggested FTP.


Stick with AI FTP for consistency, Adaptive Training will take care of the rest - the difference between a level 1 and level 10 SS workout is far greater than the difference between a 200 and 209W FTP.

I think the only reason to do a 20 minute test these days is if it’s specific to an event you’re doing. I.e. If you’re training for a 10 mile TT (though even then you’ll also want to do it in a TT position to be as specific as possible).


@AJS914, @WindWarrior , @OreoCookie , @cartsman,

Thanks for the insights, guys. This is very useful. :call_me_hand:

I have decided to follow the AI suggested FTP and move forward.



Different FTP tests produce different results. That’s the bottom line. If I understand your post correctly, AI gives you a result of 209, and you have been able to complete 100% of workouts based on the AI FTP in the past.

My advice is to continue following the AI FTP advice. So far so good.

Just to give you an idea, here are some results from the study linked in the ‘FTP is not your 1 hour power’ thread on this forum. Note the outliers for the recreational & trained groups. It varies from just over 20 minutes (for a 20 minute test!) to over 75 minutes. There is tremendous variation in TTE at 20-min-test-derived FTP. Even the pro cohort has one rider that couldn’t make it more than maybe 25 minutes (cav, prolly).



I’m interested to know why you choose to do the 20 min test if you weren’t going to use the result?

My problem with 20min FTP test is that you have to know before what your FTP is… :blush: I did it one or two times and knowing that my new FTP should be at around - let’s say - 200W i should pace it at 200/0.95=210W.

I got the ballpark from my previous FTP result and how I felt during my recent threshold workouts and a guess how the new FTP should be.

Better stick to ramp tests or/and AI FTP.


This is why I like the longer Kolie Moore test. It’s ramping and you’ll hit that spot where you know you are at your FTP or above and need to back down. It really helps you feel the spot and also gives you a decent TTE if you use something like WKO.


Hi, I have done a ramp test to find, about where my FTP is and then did the 20 minute test. It has been a while but I have tried an all 1 hour test too. I have used the 20 minute test as a workout.