Adaptive Training Is Now Out of Beta & Fully Available to All TrainerRoad Athletes!

There’s a definite “learning” of your survey responses going on under the hood from what I can tell, so I don’t think it is as simple as that. I am pretty sure Ivy or someone else from TR has directly commented that it’s not as simple as that either.

For Productive/Stretch workouts, 3 - Hard has probably been my default response, and I have never had future workouts adapted downwards from that. When I have used 2 - Moderate, I’ve had future workouts in that progression adapted upwards. So for me it “feels like” 3 - Hard is meant to be the default response for those types of workouts.

When I have ranked workouts 4 - Very Hard, I have never been presented a struggle survey, so I assume there must be something else that TR is doing to identify that it might have been a bit of a struggle (big cadence drops after intervals for example?). Or alternatively, it may simply have “learned” that this athlete never uses 4, so if they do so it means they must have really struggled?

1 Like

I’m new to AT, but I’ve only had the struggle survey for “All Out”, not when I’ve put it as Hard even where I’ve grinded away. Or like this mornings clusterf*ck where I had multiple pauses (it was just too hard).

Similar to yourself, I think it’s more nuanced than the survey. Productive Sweetspot classified as moderate, no adaptations, because I assume that’s where it’s supposed to sit in RPE. Similarly, endurance I’ve rated as Easy, have had no adaptations.

1 Like

I guess I’m the only one who gets a fail survey after selecting hard, most of my workouts up to this point I have ranked as moderate. I went back to update survey response for lazy mountain, which is basically a recovery ride. To play with the system I selected hard and It then went to the failure survey after selecting hard. I didn’t keep it ranked that way but more out of curiosity to see if the same thing would happen, which it did for me. Maybe the responses are individual based on how we historically rank them? :man_shrugging:

1 Like

I’ve rated a few SS and Threshold sessions hard but I haven’t so far had a fail survey. I could be wrong but I would think it looks more at how you actually performed in a session against the targets before issuing a pass or fail survey and only take in the response in respect of future adaptations.


That could be the case, as long as it’s consistent it doesn’t bother me. Maybe they’ll clarify what’s happening with survey responses, if they’re individual or something else. I also feel they don’t want to share too much information because people will then manipulate survey responses to give them harder workouts. Thus leading to burnout and whatnot which was why they created AT in the first place, to prevent overtraining/ burnout.

1 Like

I’ve been “successful” in that I’ve completed a workout, but rated it as All Out, and got the fail survey. Presumably as it shouldn’t have been all out RPE!

My best bet is that All Out will always give the Struggle survey.

Like others, I’ve never had that survey for any other rating. But based on what it seems TR Support added, that may be changing.

If so, it would be REALLY nice to actually know that for certain from TR.


Agree maybe @IvyAudrain can shed some light when it’s morning time on the west coast.

1 Like

I dont know if there is a standard response. I never used hard for a workout until yesterday. I received a response after marking the workout hard asking what made it hard. In my case I answered training fatigue.

To me it is just more data capture.

1 Like

Yes, but really my point was (and something I’m only learning) is to be honest in the survey results. That workout wasn’t just “very hard” based on the description, notwithstanding a reduction in intensity. Regardless of what happens back end, surely the survey result should’ve been All Out? Then complain if the adaptations down don’t happen, rather than say AT isn’t working after not being honest with the survey!

Of course, nonetheless and consistency is essential. I’ve worked hard to nail down what the survey responses really mean.

There has been more than a little confusion and ambiguity over them from the very start. I made my own definitions from the start (when all we had was a single word and number) and then refined it with communication from TR. I know have better definitions for each level and use it as such.

Interesting :thinking: Did you manage to hit your individual intervals targets ? Pre TR I’d always finish my ex-coach’s session but on the last block I may have dropped to an intensity that let me finish. Pure guess but if I did that in AT TR would label that as a fail (a failure to hit targets) whereas its a successful completion of the work out (duration wise).

Hit targets, but it was all out. So that’s what I put. Smith -2 earlier this week. (I had some connection issues, so only one pause was a genuine pause due to the workout). Actually when I went to check it there, my survey result hadn’t seemed to register, so I again went “All Out” and got the struggle survey.

This week has been tough, but that’s partly my fault, as I rebuilt with me not making my A event. Hard event last weekend, and that plan rebuild turned this week from recovery to work (with next week as recovery).

1 Like

I think a TR person said somewhere that when that happens it defaults to a fail/struggle survey and you have to over ride it with ‘I did not fail this workout - Device Issues’ but I couldn’t say for certain.

Still a bit puzzeled concerning the consequences of rides for my personal levels.

For example, these are the main rides I did the last few weeks. I nailed all of them. No pauses, no backspinning. Reached and even exceeded TSS & IF for every ride.

  • McGregor +1 (sweet spot) : “hard” > result : progression + 0.1
  • Dicks -1 (threshold) : “very hard” > result : progression + 0.5
  • Kosciuszko -2 (VO2 max) : “hard” > result : progression + 0.5
  • Antelope +3 (sweet spot) : “moderate” > result : progression + 0.7
  • Jacks +4 (threshold) : “all out (poor sleep)” > result : progression +0.5
  • Tray Mountain (sweet spot) : “hard” > result : progression 0
  • Sanford -1 (threshold) : “moderate” > result : progression 0

I want to have faith in AT and do not try to “tinker” with the results by means of altering my responses. I try to be consequent. But I don’t understand why some workouts result in a progression (for example after a “all out” one) and why some won’t yield any progression (for example after a “hard” but well executed workout).

Since I entered AT, the system only suggested adaptations once. I probably should take this as a sign that both my FTP and my plan are spot on, right?

Got a ramp test coming up next week. Curious as to what the future holds in store!

Just looking at the SS workouts.
McGregor + 1 is ranked SS 6.2. Your career SS PL’s must have been at 6.1, hence +0.1 to raise your career SS PL to 6.2.
Antelope + 3 is ranked SS 6.9 so doing it raised your career SS PL from 6.2 to 6.9 (+0.7)
However, Tray Mountain is ranked SS PL 6.9. You are already at that level (well done) and you won’t progress your career SS PL (Result Progression = 0)

Regardless of the survey responses.


Trying to explain any ML is a bit like reading tea leaves, unfortunately. What you see is not what’s happening under the hood. Also short durations—one or two datapoints—are not enough to draw any conclusions. Instead of looking at each individual workout, try to see if the overall plan and your improvements over time make sense.

I also don’t think we should expect adaptations often. It would suggest that your plan almost never makes sense and the AI is constantly trying to catch up, whatever that even means.

1 Like

Just to add my recent experience…

This morning I did a custom workout intended as a 40 minute threshold effort. I marked the effort as hard and was then given the fail survey afterwards. I answered that truthfully and submitted. I received all of the expected PL progressions

So…my guess is Hard → Fail, you get PL updates. All out → Fail, you don’t get full PL updates.

Hard to know though with the relatively small sample size

At least I’m not the only one, I think they want more information to feed the system. Makes sense to me that if it was hard, maybe there’s a reason for it. Like I said above most my workouts have felt moderate, not too hard and not super easy. Makes me think they want to give mostly moderate workouts that are tough but achievable. I’m still happy with AT it’s just a learning experience for everyone while it’s in beta.

We are definitely hearing athletes that a better understanding of what survey responses will lead to and how to effectively answer them is a common point of confusion, so the team is working to either add context with more information on the surveys or re-phrase a little to avoid further issues and frustration.

Until then, try to answer honestly to how the effort felt, and if you get a struggle survey but dont feel you should because the workout wasn’t a struggle, just answer with ‘I didnt struggle’.
However, a workout wouldnt be ‘very hard’ or ‘all out’ if it wasnt a struggle to execute, so it should be categorized accordingly as a struggle.

1 Like