For the foreseeable future, we’re using data to make incremental improvements to our training plans.
We’ve got the data!
We’ve got years and years of high quality, accurate data on how our users complete TrainerRoad workouts. We know their age, their watt/kg, how often they’re training, their history and where they are in a plan.
I’m talking about millions and millions of workouts.
We’ve leveraged some of this data and built an internal system to categorize workouts in the following categories:
- Completed Successfully
- Completed, but not totally compliant
- Failed
We’ve then developed a “Compliance Rate” metric. If you’re in group 1 or 2, we say you’re compliant.
Hard workouts should be hard
In general, our theory is that hard workouts are hard, and they shouldn’t have the same compliance rate as something like Pettit which is aerobic riding for 60 minutes. That being said, there’s also a minimum compliance rate that we’d like to hit.
We have found that some workouts in plans have a lower compliance rate that we’d like.
Sometimes these are because the workout is too soon after a Ramp Test. Other times small tweaks (like increasing RBI from 2 minutes to 2 minutes and 30 seconds) reduces the failure rate by half and doesn’t change the nature of the workout.
With this data in hand, Coach Chad is making small tweaks to our plans.
Plan TSS changes
One thing to note is that TSS could change when we make these tweaks. Some people are OBSESSED with TSS, and if they don’t have an exact point load per week, they will freak. TSS is helpful, but it isn’t everything. We can get the adaptions we want and make you faster even if TSS isn’t a perfect point step.
Timeline for updates
We’ve already updated Sweet Spot Base I/II Mid and Low Volumes.
Next up are our build plans. We’ll continue this process for (hopefully) ever. As we get more and more data, we’ll make smaller and smaller changes and customize it more for each individual.
What if I’m currently on a plan?
If you’re currently in a plan, there’s no need to switch your workouts with our new ones, but you are welcome to do that if you want. Think of these as 2-5% plan improvements rather than the new “hot shit.”
If you want to be aware of updates, please watch this thread and the forum. We’ll post when we make updates.
What’s your exact process for taking into account a failed workout?
We won’t be sharing our exact process to figure this out. It was challenging to develop, and we view it as a competitive advantage.
I can tell you that we’ve thought about people ending workouts because they are busy, equipment failures, skipping just one interval, etc.
If you have any other questions about the process, let me know in our forum post about this topic!
Ready to get faster?
New to TrainerRoad or ready to re-subscribe? We’ve got amazing training plans, a training calendar, cycling analytics and software to make you a faster cyclist! Learn more or re-subscribe at www.trainerroad.com.
-Nate Pearson
CEO & Co-Founder
Share this Post
As a person who has built a company and spent a decade on web analytics, I’ve spent many hours trying to decipher “why did a person do what they just did?”, so I applaud your efforts. However, one suggestion I have is simply to *ask* them. When you end a workout early, have a screen that says “why did you stop this workout? And have little boxes to hit with “I’m short on time today”, “I’m feeling sicks”, “fatigued from previous workout”, “it was just too hard”, etc.
You’ll get much better, cleaner data than trying to guess and this clean data will actually improve your other algorithms as well.
That is coming in the future. I think we’ll want both ways. And we have a lot of historical data without any responses.
Sorry, yes, I should have been more clear what I meant: the user specified input will make your algorithms that process the non-response data more accurate since you can train it against them (i.e. our algorithm thinks you ended this early because of fatigue–then check the response to see if it was right). So you can definitely have it both ways and even the non-response data will gain from it.
if I´m on a plan and want to have the new workouts, what do I have to do?
Unfortunately there is no way to automatically switch over to the new plan 🙁 For now, you will have to compare your upcoming workouts vs the workouts in the training plan.
For future changes, we’ll make it more clear about what changed.
This is awesome! Personally I really do try and follow the workout plans closely and it is a morale boost everytime I finish them clean….but just barely holding on. Conversely when I have to backspin several times before the last interval, or bailout completely I have a drop in morale (it will still happen I am sure but the frequency may reduce). Using data analytics to tweak plans……very, very well done to all of you at TR. Thanks for the continuous improvement.