Trainerroad makes you a faster cyclist, but how much faster? From the thread ‘The Bell curve of cyclists’ we already know how fast the average TR cyclist is. But what about the average improvements realized using Trainerroad? The thread ‘Your FTP improvements’ has 1098 comments as of today, and answers this question for individual users. Here a best-effort exploration of the aggregate data in that thread.
Of the 1098 comments posted, 185 self-reported results could be retrieved that included 1) starting FTP, 2) plan and plan volume, and 3) resulting FTP. This data-set might include some doubles as many members posted updates, and certainly does not include all results posted as some were hidden in longer posts. Posts that mentioned re-calibration of power meters, intermittent illness or doubts about FTP test accuracy were ignored. This resulted in a raw data-set, which contains no information about age, gender, weight, prior training history, general health, consistency of training, etc, etc.
The average user in the forum thread had a starting FTP of 235 and reported a 17-Watt improvement after a (any) single training plan phase, or a 7.9% gain.
What happens when fitness improves, what is the difference between low-, mid-. and high-volume plans, and which specific training plan phase results in the biggest gains? Plotting all 185 data points it becomes apparent that improvements are getting smaller as starting FTP increases, as expected.
More than 90% of self-reported results were from low- and mid-volume plans (evenly distributed), the remainder was on a high-volume plan. Separating these out leads to three different linear correlations (R-squared 0.52, 0.34 and 0.65):
The choice of a linear correlation can certainly be debated but keeps the logic simple. The low number of data points for the high-volume plans is a limitation, but with the correlations above one can calculate an average expected gain from a (generic) plan as function of starting FTP. So what is the influence of the type of plan? Here the size of the data-set is again a limitation, but for low-volume plans it looks like this:
From the limited data one can conclude that none of the plans leads to better results across the full range of users, and for the average TR user (~230W, ~3.0FTP/kg) the difference is not substantial. Remember all the blog posts stating: ‘the real improvements will come after the base phase’? This limited data-set does not support such conclusion.
Finally, what FTP gains could an average TR user expect from a number of consecutive weeks of dedicated training? Using the linear correlations from the low-, mid-, and high-volume plans (and ignoring plan-phase-specific influences) one can get a rough estimate. As not all plans have the same duration this was (arbitrarily) set to seven weeks. And as data points for mid- and high-volume plans are not available for lower starting FTPs they are not plotted in that range.
The (invisible) error bars on this plot are huge, the data is probably biased (stronger incentive to self-report positive results), and FTP is a concept with limitations. But still interesting, and safe to say that Trainerroad will make the vast majority of us a lot faster!
Update
The results above only apply to the average TR cyclist. For individual TR users the question still is how age, build and gender factor in when predicting FTP improvements. The raw data does not answer those questions, but this blog by Joe Friel (Author of the Cyclist’s training bible) does provide more insight and is surprisingly consistent with the data above. It also turns out that the linear correlation between reported FTP percentage gains and starting FTP is a first-order nonlinear differential equation that can be solved analytically 
. Using the analytical solution and Friel’s rule-of-thumb to scale the parameters, one can predict (average) FTP gains more specifically for individuals based on their age, gender, build, current FTP and plan volume.
Link to the Spreadsheet below (again, best effort), hope it provides some additional insight!
