The problems with the UX, data and info scattered all over the place as you say, that problem has to be obvious to the people behind Xert. It truly is an example of Rube Goldberg in action. Todayâs feature addition adds to the UX clutter. I think thatâs how the UX got the way it is, one layer at a time. Rather than a revamping (expensive), you simply add to whatâs there in the UX. After a few years, the present UX is what youâre left with.
1 Like
Your criticism of TR still sounds fanboyish. Iâll leave it at that.
Along with challenges come rewards, but for some, the safer strategy is to allow those with deeper pockets to lead the way. I get it.
Lol - when lacking arguments⌠throw stones.
Thank you for reinforce my point of how TR uses are moved by passion not rational.
1 Like
there are a few very specific items that I think make Xert unique, one of the most prominent in my mind is Smart Intervals. Most of the programs out there still use a single factor for training prescription, some might be trying to use more. I agree with Armando on machine learning, and in general multivariate modelling and prediction. Though in traditional multivariate modelling you can see where the levers are, NN and other machine learning programs tend to make it very hard to understand what is being used to predict what. Also it really takes the garbage in and out principle to the extreme. I have worked with multivariate modelling and can by no means be considered any type of expert on it. However though it can be very useful one needs to be careful that you do not let the machine run your world, you still need to use critical thinking.
What I hope for and why I pay for Xert is that it will grow and mature and find its feet. Remember, just because you are better does not guarantee a successful future, you still need to listen to your customers, ask Computrainer. So please take the time to listen, learn, and adapt. Your customers are your business not the Xert program.
3 Likes
Hereâs a relevant conversation from 2019. There are some Xert comments too. https://forum.slowtwitch.com/forum/Slowtwitch_Forums_C1/Triathlon_Forum_F1/A_New_Approach_To_Predict_Performance_P6836199/?page=-1
Thank you Ron for that valuable feedback.
There is a lot of growth needed for sure but not just from us but also from the community of athletes, coaches and other content providers. There is still a lot of inertia around using FTP, FTP testing, old style training plans. Just go through the GCN library of videos on power and see how many refer to FTP vs. how many refer to a more than just FTP.
There is a lot of work to tack this ship and in many ways itâs independant of the money we can throw at it. People are just resistant to change and with Xert everything changes. We need the community at large to get it, reach new levels with it, win with it, share their experiences and for all those, it will take time.
Weâre always watching and reading everything our customers say but we also track our data to see how customers are really responding. Customers complain about the UI/UX but a lot of that is driven by the newness of the methods and the complexity of what is happening on the system. Weâre always trying to strike the right balance between creating abstractions for simpfications with the desire for more information and control. Customers complain on both ends so always a challenge! 
This is made more challenging given everything is new and much of what weâre doing has never been done before⌠not like we have a lot of good business models to base things on. We try something out and see how customers respond. Fortunately we havenât yet had to re-engineer too much of the system. At least not recently. But lots of things are happening around here. No rest for the weary!
Personally, I like the fitness modeling and how Xert implements it. Iâd like to do more of that and give athletes (and myself!) an even better representation of themselves as athletes, how they perform and how they can improve. We have changes planned likely affect a number of things on the system, someways simplifying things but also making it much more difficult on the back end (as if it wasnât already!!!).
One of the things we think will really help is a better onboarding process for new users, especially those that are new to power. We donât really have a good process for them and we could use something that will step people through the system, the concepts and how they can use it, train with it, using it on rides and ultimately to reach new levels of fitness. Itâs all there at the moment but just a bit harder to work if you donât have a relatively deep understanding of it. In some ways, new users have an easier time with Xert than experienced users do since they donât come with expectations on how the software should work. Users that come from other systems expect to see features they are accustomed to and havenât yet figured out that in many cases, those older ways of doing things are obsolete and they can find improved ways of doing things on Xert ⌠but theyâll need to invest some time to figure that out. Some arenât interested in that much change so will complain about lack of this or that or about something on the system - âXert underestimated my FTP. Didnât work for me!â
We tend to focus mostly on essential pieces athletes need to discover, improve and perform. Other systems may offer users various tools to slice-and-dice information or apply advanced filters from a massive library of workouts or plan an athleteâs training plan for the next 6 months. We donât deem these essential so if youâre looking for these, by all means use Xert with another system. Remember, weâre only US$9.99/month. If you simply want to track and improve your fitness, even reaching your best fitness ever, there is certainly enough on Xert for you to give up everything else and just use Xert. Weâre seeing a lot of users doing just that, especially now with Xert Sessions. We do however take in what users ask for but we balance these requests with what will make the most impact to users and their overall fitness. Thatâs the bearing we keep.
Keep the comments flowing! Always happy to interact with customers, even those that arenât so enthusiastic about Xert as we are.
3 Likes
Armando, I see the progress and understand how hard it is to navigate the waters in an uncharted sea of information and in particular the exercise space. One thing I do feel you have right and that is getting away from the thing called FTP, so few people understand it and worse consider it a value statement or something that imparts value to them rather than just like any measurement it is nothing more than a metric to help determine where you are and where you are going⌠it is personal but only of value to you and should not be used to compare against anyone else.
I truly only want Xert to get better. otherwise I would not waste my time writing comments. The fact that you engage with the users is a very good start!. Cheers Ron.
Yes I am familiar with that thread and Alan Couzens feelings about data reductions/ modelling. I may and may not agree with all that either Alan says or Armando but it is a deep topic and I probably donât have the depth to do the math to reinforce either argument. My background is other uses of multivariate modelling and I have used older NN programs as well but in the end the utility of any model is how well it predicts reality. When Xert is set right my experience is it does that pretty well. That is not to say that other models arenât as good or better and maybe even Xert can get better with some as not used variable. What I tend to not like on ST is the personal attacks. Argue about ideas, etc but it need not get personal.
From what I can see Xert is just a novel way of presenting and allowing users to use the data, one that is based on a legacy Bannister/TSS model of training stress. The workflow method is presented in a unique way, feedback from one unscheduled workout at a time, but itâs still built on top of a legacy Bannister system. You can do similar by simply looking at a PMC (Performance management chart) after each workout, see where you stand, and then decide on the next dayâs workout based on that readout. Anyone who has experience with TSS, CTL and ATL on a PMC chart experiences the limitations of this (and Xertâs) method. A personâs form (TSB) can be high, and yet he can in reality be tired. Same in the other direction. A CTL approach can show one as being tired, and yet the athlete can be ready to win races. Thatâs why most serious athletes supplement that legacy approach with a human beingâs neural net, better known as a coach. Xert allows one to use a limited neural net, oneâs own, and hence the use of the freshness/feedback slider. The FTP estimation (something pitched as a big feature by Xert) is interesting. Strava does the same. So does Intervals.icu. I think they use different methods, but I doubt Xert is doing anything novel. Explanations of Xerts methods by management are often littered with abstruse, proprietary jargon and pufferyâŚa big turnoff. Same abstruse, jargon-filled puffery with XATA. Intervals.icu tells you what method is being used, Xert and Strava donât. SoâŚif Xert works for you, then great. As I wrote earlier, the user interface is a complete mess, and I took a guess as to why in the earlier post. Armando has said itâs a resource issue in previous postsâŚThe management decision is that it makes more sense to use the limited resources in other areas. The rah-rah from some users (and management,) the cult-like behavior I find off-putting. Enthusiasm is good. Hype and puffery is not. On the positiveâŚXert is fun to use. Once you get a handle on the screwy interface, itâs fun to get a âbreakthrough.â Iâm not sure how accurate these breakthroughs areâŚbut the user can judge that for himself. One thing Iâm liking about TRâs new thingâŚAs they say, the best measure of performance is performance itself. Talk about FTP being dead? I think thatâs the way TR is going. The workouts suggested are being adjusted (partly) by how well you can complete them, with ongoing adjustments. Thereâs no need to know oneâs FTP number.
My view is that from looking at how TR seems to be struggling with a fairly messy implementation of adaptive training I have a whole new appreciation for the more elegant maths that Xert is built on.
But yeah the Xert interface looks like itâs been done by engineers because presumably it has been.
2 Likes
TRâs Adaptive Training is in beta. As far as Xertâs design, If you will admit that Xert has been designed by a protege of an engineer named Rube Goldberg, then you and I are in complete agreement!
It is quite odd really. On TR there are threads proclaiming Xert as the best tool and the opposite happens on here.
FWIW I love Xert and I can totally rely on my TP being very close to accurate. Do wish I knew more about it though. I listen to the podcasts and read the forums and do have a general understanding but certainly not as deep as I would like.
1 Like
Youâve fixed it for me already. You guys are amazing. Thanks for the fix AND for the exciting new feature.
1 Like
Weâve also fixed some of the calculations and improved the projections too. For Challenge/Continuous projections, rather than expecting your training to continue as before, we now make the assumption youâre going to train as the Athlete Type Focus with Mixed Specificity over the period. You 'll notice subtle changes in the projections when you change athlete types as a result. Cool?
2 Likes
Cool indeed 
1 Like
Youâre mirroring. Here are some facts. That TR questionnaire is just one data point. You/we have no idea what other data points TR is looking at, but if you take a look at the science, you can see that the potential is enormous. ML and neural nets are the future. Xert on the other hand is built on a 50 yrd old theory, the Bannister impulse model. Xert is a novel but rather obvious workflow, one that metes out the data that is being generated by that 50 yr old theory. If youâre interested, read up on the shortcomings of the Bannister model.
What is the point in coming on a Xert user forum to tell us how outdated it is?
If you donât rate Xert then donât subscribe.
4 Likes
⌠and the reality is that it isnât outdated. Itâs just that using other measures to drive it donât work and those are the documented shortcomings. The other shortcoming is that it needs regular feedback (fitness testing). Both of these are addressed with Xert where we use XSS and Fitness Breakthroughs to drive the model and we split it into 3 concurrent models. Again dealing with a shortcoming of the model - more than 1 measure of training dose. Doing all these, you end up with a pretty darn accurate model. For example, check this out:
6 months of dead reckoning using XSS and No Decay. Who needs an ML model? Note that weâre not just predicting his signature even. Weâre predicting the moment-of-failure to within seconds! Thatâs based on 6 months worth of training data. ML will never be able to do that. You donât have moments-of-failure identified in the first place!
(Carmen) As with most of your posts, youâre misinformed and misrepresenting, often provocatively. I would kindly ask that you refrain from doing so. Like @johnnybike says, youâre welcome to do so elsewhere but provoking arguments through misrepresentation isnât what our users want out of other users on this forum.
13 Likes
I think we should not go down the road which method is âbetterâ as both programs (tr and xert) have their outstanding features (one they have in common is the two passionate ceos) and shortcomings. In the end you can get to your max capability (as an amateur/enthusiast with ambitions; upper end of that group as I personally define it: 4.5ish w/kg) with both (or neither 
) programs. You just have to put the frickin hours/quality in. Iâm under the impression that some people search for âthe bestâ program as a shortcut. Tr, xert, suf, you name it. In the end it comes down to personal preference. For me it is xert as it allows me to understand my physiology better. Machine learning canât do this, as it is by its nature a black box. But every (good) coach will tell you why you have to do what session.
As for the xert shortcomings ( I m too lazy to open a new threadâŚ) : could you shortly explain (or point me to an article) why I wonât have a breakthrough at 1min on 1min off vo2max repeats? Map goes up during the 1min off to max again, so I could theoretically go forever (which I obviously canât 
) @xertedbrain
2 Likes