By Dr David Bell, Specialist Anaesthetist (Retired), Software Engineer, and Founder of Align AI Fitness, NSW, Australia
In Part 1 of this series, I covered 25 years of research showing that tempo, not genre, is the primary driver of music's ergogenic effect during exercise. The evidence is clear: match the beat to the intensity, keep it between 125 and 140 BPM for most work, and the genre takes care of itself.
So here is the obvious question. Do the platforms that millions of people use every day actually do this?
I looked at how five major fitness platforms handle music: Peloton, Apple Fitness+, Les Mills, Spotify, and Beachbody (now BODi). The answer ranges from "surprisingly well" to "they are not even trying."
Peloton: Great Music, No Science
Peloton treats music as entertainment, not as a training variable. And they are very good at the entertainment part.
The company has a dedicated Head of Music, five music supervisors assigned to individual instructors, and a proprietary internal tool called Crescendo that lets instructors search over a million licensed songs by BPM and duration. In 2018, they acquired Neurotic Media, a B2B music streaming startup, to build this capability. Goldman Sachs estimated Peloton contributed roughly $267 million to the music industry in 2022, paying approximately 3.1 cents per stream (compared to Spotify's 0.35 cents).
But the critical detail is this: instructors choose their own music. Paul DeGooyer, Peloton's former Head of Music, said it directly: "Absolutely not. They choose their own. We don't promote anything, musically speaking." The five music supervisors support but do not dictate.
This means your Peloton playlist is a function of your instructor's taste, not exercise science. Some instructors function essentially as DJs. The energy is fantastic, but whether the tempo matches the prescribed intensity of the interval you are doing is largely a matter of chance.
They also learned the hard way that music licensing is not optional. In 2019, the National Music Publishers' Association sued Peloton for $150 million over unlicensed songs. A federal judge later allowed the complaint to be amended, doubling damages to $300 million and covering 2,468 works. Peloton pulled the infringing songs, wiping nearly half their on-demand library in what users called "The Purge." The settlement cost $49.3 million. The music is legitimate now. Whether it is optimally programmed is a different question.
Apple Fitness+: Polished, Intuitive, and Unscientific
Apple's approach is more curated than Peloton's but still relies on human intuition rather than data.
Trainers choose the musical theme as a foundational step when designing a workout, then work with Apple's in-house music specialists to build playlists that match the mood and structure of each class. Apple hired trainers partly for their music industry awareness, not just their fitness credentials. The integration with Apple Music is seamless: users can filter workouts by genre, specific songs, or artists, and an Audio Focus feature lets you adjust the balance between trainer voice and music.
But there is no evidence of systematic BPM-to-intensity matching. The tempo selection appears to be a manual, intuitive process. That does not mean it is bad. Experienced trainers develop an instinct for pacing, and the music specialists add a layer of quality control. But instinct is not the same as the Brunel Music Rating Inventory, and "this feels right" is not the same as "this is 128 BPM at 65% of VO2max."
Apple has the data infrastructure to do this properly. They have real-time heart rate from Apple Watch, cadence data from accelerometers, and the entire Apple Music catalogue with audio feature metadata. The fact that they have not connected these dots is a missed opportunity.
Les Mills: The One That Actually Uses the Research
Les Mills is the exception, and it is not close.
The company has been fusing music with structured exercise programming for over 50 years. A dedicated music team led by Creative Director Diana Mills listens to thousands of songs every month to select the 10 to 12 tracks per new workout release, producing 80 unique playlists per year across their 20 programmes.
What separates Les Mills from every other platform is that their approach is music-driven: workout choreography is built around the music, not the other way around. For a BODYJAM release with 20 songs, choreographers will apply to license 60 to 70, with roughly half denied immediately. The music comes first. The movement follows.
More importantly, Les Mills explicitly references the exercise science. They cite Professor Peter Terry's research on the psychophysical effects of music. Their guidelines specify that warm-up music should sit at 100 to 120 BPM to match target heart rate elevation to roughly 110 BPM. They reference research on synchronous music lifting enjoyment and performance. They even cite neuroscience findings about the left inferior frontal gyrus and how the combination of music and exercise modulates the perception of exertion.
This is what it looks like when a platform takes the research seriously. The result is a structured, repeatable, evidence-informed process. It is also, not coincidentally, the reason Les Mills classes feel different from a random Spotify playlist on shuffle.
Spotify: They Had It, Then Killed It
This one is the most frustrating.
In May 2015, Spotify launched a running feature that used the phone's accelerometer and gyroscope to detect your running cadence in real time, then automatically matched songs from the catalogue to your detected tempo. It supported a cadence range of 140 to 190 steps per minute. This was, in principle, exactly what the research on entrainment and synchronisation suggests should work. Van Dyck and colleagues showed that runners spontaneously adjust their stride rate to match music tempo with an r-squared of 0.96. Spotify had built the tool to exploit that mechanism.
They retired it on 26 February 2018, citing that the feature was "vastly underused."
Today, Spotify offers no native real-time tempo matching. Users must manually search for BPM-specific playlists. Spotify does not display BPM on tracks. And a reported audit of 500 top workout tracks found that 68% of Spotify's listed BPM values deviated by 6 or more BPM from ground-truth audio analysis. That is a problem, because the entrainment window is narrow: it works best within plus or minus 1 to 2 percent of your natural cadence and drops off sharply beyond 2.5 percent. A 6 BPM error is well outside that window.
Third-party apps like Running Beats, PaceDJ, and RockMyRun have tried to fill the gap, but they require leaving the Spotify ecosystem and introduce their own reliability issues. The most scientifically interesting feature Spotify ever built is gone because not enough people used it. Which says more about product design than it does about the science.
BODi: Vibes Only
Beachbody (rebranded as BODi) uses a trainer-driven model where "Super Trainers" personally curate playlists based on taste, recommendations from friends, and Shazam discoveries. There is a BODi Spotify channel with programme-specific playlists. There is no evidence of systematic BPM matching, exercise science-based selection, or any structured process at all.
I am not going to spend more words on this than the platform spends on its music programming.
The Gap Between the Research and the Product
The pattern across these platforms is consistent. The research says tempo is the primary ergogenic variable. Karageorghis's work shows a ceiling effect above roughly 145 BPM, where motivational gains flatten even though many spinning and HIIT tracks sit at 160 BPM or higher. The optimal range for most exercise is 125 to 140 BPM. Entrainment (your body unconsciously syncing movement to the beat) works within a narrow tempo window and requires accurate BPM data to exploit.
And yet: Peloton lets instructors DJ based on personal taste. Apple Fitness+ relies on intuition backed by music specialists. Spotify killed their one science-aligned feature. BODi does not appear to have a process.
Les Mills is the only major platform that explicitly builds on published exercise science research. They are also the oldest, the least "tech," and the one least likely to show up in a venture capital pitch deck. There is probably a lesson in that.
What This Means If You Train With a Platform
If you use Peloton or Apple Fitness+, the music will be good. The energy will be right. But the tempo may or may not match the exercise prescription for the interval you are doing. You are getting entertainment value, which is not nothing (enjoyment drives adherence, and adherence is everything), but you are leaving ergogenic benefit on the table.
If you use Les Mills, you are getting the closest thing to evidence-based music programming that exists at scale.
If you use Spotify for your own playlists, you are on your own. Check the actual BPM of your tracks (tap tempo apps are free and more reliable than Spotify's metadata) and build your playlists around the tempo targets from Part 1.
And if any platform is listening: you have the data. Heart rate, cadence, power output, movement patterns, and the entire history of music psychology research. The feature that would actually connect these is not technically difficult. Spotify proved that in 2015. Someone just needs to build it and keep it running.
This is Part 2 of an 8-part series on music and exercise performance. Next: the neuroscience of why music reduces perceived exertion, and what happens in your brain when the beat drops mid-set.
Dr David Bell is a specialist anaesthetist (retired), software engineer, and founder of Align AI Fitness, based in NSW, Australia.