Has Adaptive Training Plateaued My Progress? A Data-Backed Review (With Help From AI)

Training
1

Hey everyone,

I wanted to share a long-term perspective on my experience with TrainerRoad — especially in relation to the introduction of Adaptive Training (AT) back in 2021. I’ve been a user since 2014 and have built up a pretty deep FTP history along the way. TrainerRoad has been the backbone of my training for over a decade, and I genuinely credit it for much of my growth as a cyclist. I regularly recommend it to friends and fellow riders. That said, I think it’s fair — and constructive — to critically assess how some of the recent changes may be impacting long-time users who’ve been through many seasons of structured training.

:bar_chart: TL;DR – My Theory

  • TrainerRoad worked brilliantly for me until 2021, when Adaptive Training was introduced.
  • From 2014 to 2020, my FTP steadily rose from ~200W to a peak of 326W.
  • After 2021, I experienced a plateau and slow decline, despite consistent use of TR and Adaptive Training.
  • Since reverting to the older-style SSB HV plans in February 2025, I’ve seen renewed gains — up to 316W in June 2025.

:brain: About Adaptive Training

Adaptive Training and AI absolutely have their place — they just haven’t been a good fit for me since I plateaued. On the topic of pushing through plateaus, I remember a great old TR podcast that covered this really well, linked here.

Right now, I’m following the old/original SSB High Volume plans — the ones no longer officially available — and they’re working much better for me.

:chart_increasing: FTP Trend Analysis

I reviewed my entire FTP history by grouping results into three phases:

Period Avg FTP Peak FTP Comments
Pre-Adaptive (before Mar 2021) ~307W 326W Strong upward trend
Adaptive Training (Mar 2021 – Jan 2025) ~305W 318W Plateau and slight decline
Reverted Plans (Feb 2025+) 312W 316W FTP rising again

From March 2021 through early 2025, my FTP flatlined. Since switching back to traditional TR plans without Adaptive logic, I’ve been steadily improving again.

:red_triangle_pointed_down: Dumbing Down of SSB Plans?

Another thing that happened around the same time as AT’s rollout was the restructuring of the Sweet Spot Base plans.

Let’s call it the “Dylan Johnson effect”. His video critiquing the intensity of TrainerRoad’s traditional Base plans gained a lot of attention. Shortly after, TR released plan updates that emphasized sustainability and reduced early-week intensities — likely a move toward improving compliance.

That’s understandable, but here’s the problem for long-term users:

The old SSB High Volume plans provided a significantly greater training stimulus.
For example:
:fire: Final week of SSB HV II = TSS 731
:bicycle: Final week of General Base HV II = TSS 591

That’s a 19% drop in maximum weekly load.

There’s clearly a balance to be had — we want people completing workouts and avoiding burnout. But if plans are too easy or conservative, many of us simply won’t see gains.

:toolbox: Request: Create A True Custom Plan Builder

TrainerRoad has said the old plans aren’t coming back. I get that.

But for athletes like me who’ve already been through years of training cycles, I’d love to see a true custom plan builder:

  • Build your own multi-week plan
  • Include full calendar logic (progressions, recovery weeks, etc.)
  • Save and reuse plans

Right now, building a long-term custom plan means manually copying workouts and fiddling with calendar weeks — it’s cumbersome and prone to errors.

:man_raising_hand: Final Thoughts

  • AT is great for newer or inconsistent athletes, and the idea behind it is solid.
  • But for experienced riders chasing marginal gains, it may fall short by capping stimulus or overly smoothing progressions.
  • I’d love to see TR support both ends of the spectrum: those new to structured training, and those of us who know exactly what we need.

Thanks for reading. Would love to hear from anyone else who’s taken similar steps — or disagrees! Always happy to exchange ideas backed by data.

P.S. Shoutout to AI for helping put this all together — otherwise this post would still be in my head, not the forum.

22 Likes
2

Not surprising in my opinion. That was a hard plan. Sounds like it was effective for you, probably not good for the masses. And I think that’s the core issue.

As adaptive was rolled out, TR has evolved into a much more conservative/safe training approach (in my opinion). Smarter, but safer. I’d hear it all the time about lack of compliance or people getting burnt out on the old plans, so it seems like a logical direction to take. If the goal is to make the largest number of people faster, you have to create a system that values consistency/compliance over all else. And while consistency might be the top priority of any training approach, I think it can also result in a system that has a soft hand and errs on the side of making sure every session is a good session. It’s a really good/smart approach for folks starting out where you don’t want them frustrated and they will get faster if they keep showing up for the workouts. And for more advanced athletes, progression levels and AT should be able to challenge them appropriately. I think the concept is solid, I just think the entire system is biased toward the less experienced athletes and time crunched athletes. And if it’s optimized for those folks, there is probably a little less thought and investment made into optimizing for athletes who have more time to train and/or want to squeeze every bit of performance out of their training (and accept the risk of burn out or overtraining when training closer to the edge). The TR system is not going to let you fly too close to the sun (which is good and bad). To truly optimize an individual’s training, you have to test/find limits, which means some struggle and failure (in my opinion). I don’t see any of that as a knock on TR, I just see it as a pretty smart approach to training that is going to work well for most. And probably work better than the majority of human coaches who charge much more.

10 Likes
3

Can you give us a little more background like age, how long in endurance sport, how busy the rest of your life is, and stuff like that?

I think if you move the slider from balanced to aggressive (or whatever it is) you’ll get tougher workouts but what if a person had 12 hours or more to train a week?

Joe

2 Likes
4

Hey @Joe, I’m 51 and 83kg. Never properly trained endurance until signing up to TrainerRoad. I’ve a busy life which is another reason I’m attracted to the old SSB plans. Limited with training opportunities so SSB gives me the best bang for buck. They’re definitely tough but I do seem to be able to recover well enough to maintain consistency and compliance.

1 Like
5

Out of curiosity: how did you manage to find the old SSB plans?

6

@dmalanda the TrainerRoad support folk were kind enough to share images of the old-school (pre-2021) SSB High Volume plans. Attached here for anyone that may like to embark on the suffering.

TR SSB HV I
TR SSB HV I1244×1209 59.3 KB

TR SSB HV II
TR SSB HV II1277×1202 61.7 KB

The milder versions (post 2021) are still available on the website (for the time being), just not linked from the main plan pages.

Sweet Spot Base - High Volume I
Sweet Spot Base - High Volume II

8 Likes
7

I find the same where TR doesnt seem to handle people with more time than about 10 hours well… If you have 15+ Hrs a week to train, you really do not need more than 2 intensity days, 3 if you can recover well, and the focus should be on getting you to do more “fun” long Z2 rides even during build periods. Moving the slider to “aggressive” adds a silly amount of intensity that would burn anyone out fast.

8 Likes
8

Just choose the “Masters” option and TR will only give you 2 hard workout per week.
@JayUnderscore

4 Likes
9

Ouch! And they’re 6-week blocks. I wonder if part of their effectiveness (for those who could keep compliance) was fewer interspersed recovery weeks. :thinking: I’ve been reflecting on whether I really needed recovery weeks this year when doing the various phases of Traditional Base. But, :person_shrugging: I’m fairly happy with where I’m at so far.

3 Likes
10

I’d be curious how your CTL compares over the years. I would expect that with the introduction of adaptive training your CTL went down. Then with the re-introduction of sweet spot base your CTL and weekly load is back up.

3 Likes
11

@Dark-Passenger, correct on both counts. With Adaptive Training my CTL was lower. Especially when I adhered to the guidance of the Red/Green Light guidance.

In my experience, the combination of AT and the Red/Green Light system (when I adhered to it) made it difficult to maintain the same training load (CTL) I was able to achieve with the older plans.

What is interesting is that now that I’ve done about four months of solid SSB the algorithm is no longer flagging me for Red\Amber days. Maybe it’s finally worked out that I can handle the load and recovery? :man_shrugging:

2 Likes
12

It appears as though ftp is all that you are considering. What about 5’, 1’ power? Durability? Not saying youre not right, but ftp only a few watts down isnt the end of the world if your 5 min power or repeatability has gotten better. And with training that long, how do you feel? Do you feel fit? Also, age does become a factor and you are going to start to decline sometime.

Just some things to consider. My ftp has stay roughly the same the last 5 or so years and certainly decrease from 10+ years ago…but im 51 now, so im fighting time.

3 Likes
13

@BillyWaldman 51 mate, we’re in our prime!

I felt markedly slower and the numbers showed on my outdoor rides, events and segments. I’m now feeling fitter, stronger and more durable.

One thing I’m enjoying as my form improves is using the Compare Activities facility that invervals.icu provides. Comparing the same TrainerRoad workout (and associated segments) over the last six months paints an interesting picture (as below). It’s very exciting to see power numbers going up again while heart rate is going down.

Wright Peak -2 comparison on intervals.icu

image
image1920×987 168 KB

Just noticed I can add Decoupling to the comparison also. The orange blip is due to non-compliance. :sweat_smile:

5 Likes
14

100% that’s what’s happening. Same for me: RLGL is a lot less sensitive for me now than it was at the start of the year, thanks to a sustained increase in volume. Also, TR had a reputation for burning people out so I reckon RLGL & AT were kept on the conservative side to rein in the people trying to do five workouts a week plus all the local group rides. But that’s part of being human, we don’t generally train to be good at training; we train to give our mates a floggin’. :laughing:

5 Likes
15

You’ve fallen into the common fallacy of taking a very personal experience and claiming it applies to the majority.

You believe that you perform better on 6 hard workouts a week. Good for you. I don’t think the majority of athletes do. In fact most coaches would limit people to 2-3 intense days per week, regardless of volume.

The other common fallacy is that the numbers (at least the ones you showed) are a significant difference when they’re not.

Your post title says you used AI. Did AI do your data analysis? Because you can’t trust AI to do even basic math correctly and consistently. And the numbers you presented are not enough to draw conclusions from. Average FTP means next to nothing and your peak FTP was lower. And the difference between phases 2 and 3 are way too small to declare a course winner.

It would be much better to show your PD and fitness curves from the same part of your indoor training season across several seasons.

6 Likes
16

What TR says - “Adaptive Training. Your abilities are dynamic and unique, and now your training is too. Adaptive Training uses the power of machine learning to continually adjust around your performance, goals, and needs. Because an individual approach is the key to getting faster.”

I think the OP is saying AT isn’t meeting his personal needs, while TR claims that is exactly what it does. Maybe it works better for others in the thick of the bell curve.

My question to the OP, why use TR at all in your situation? Make your own plans in Intervals. Adjust and learn on what works best for you. That is the direction I chose as I agree that TR’s programming, Adaptive or not was not working for me.

2 Likes
17

I value the workout/plan library and my training history. There’s no doubt I have TrainerRoad to thank for the huge fitness gains I obtained until I reached my plateau. I’m sure it can continue to help just not with the default easy button. I’m having to go down the customization path. Would be amazing if they catered for this model a bit better than they do today.

1 Like
18

But they haven’t shown that it’s not meeting his needs. Not that the old plan is indeed working better.

They also apparently haven’t played with the setting for how aggressive to be with training. They need to do that if they want more training load.

19

At 51 getting within 3% of your best is fantastic…and it seems like you might get even closer or possibly surpass it.

My ftp fell off of a cliff in my late 40’s but I’ve manage to keep it relatively stables since then (330-340-ish in my late 40’s down to 297 at my last AI FTP and ramped 299 (last week)). The other thing is that even sweet spot mid volume just killed me, I bombed out half a dozen times. For me, the 2 interval days and a long one or two seems to be working, I’ve been excited to train for the last year and the long rides are in the 4-5 hour range so my volume is slowly coming up.

I’m thinking you may be on the “can really recover well” side of the curve along with “stays fast as he gets older”…a potent combination! Maybe “needs a lot of volume to maximize fitness” too.

With 10 years of data…I wonder what the AI explanation is for your case? I mean…it’s had the same data you’ve had yet you’ve been getting better results with a different plan than it’s been recommending. I’d love to see something like this on the podcast.

Joe

2 Likes
20

Thanks for the kind words @Joe. I’ve was also curious so reached out to support and had a lengthy exchange with them. As you’d expect they held the line that AT, RL/GL and AI was the best approach to take. Given my data analysis I didn’t want to waste another year of pootling along on a declining plateau so embarked on my current journey of what had worked well for me previously. I did also e-mail the Podcast team on the chance that they’d offer some more insights.

4 Likes