Rethinking What Power Meters Mean for Runners
A leading company redefines what it's actually measuring, and explains why that's what we really wanted all along
Outside's long reads email newsletter features our strongest writing, most ambitious reporting, and award-winning storytelling about the outdoors. Sign up today.
Stryd, the company that pioneered the idea of power meters for running, recently published a scientific white paper called “Running Power Definition and Utility.” That might seem like an odd topic for a company that’s been selling power meters since 2015. You’d figure they must know by now what running power is and why it’s useful.
But these questions are far knottier than you might think, and Stryd has always been fairly forthright about admitting this. In Outside’s initial coverage of their launch, one co-founder said their fundamental challenge was “lack of knowledge,” and hoped initial users would help the company figure out what its product was good for. In the years since then, Stryd has gotten excellent word-of-mouth. The users I’ve spoken to have found it helpful. But there’s been a nagging disconnect between the positive user reviews and the general consensus of scientists who actually study running, which is that “running power” is a fundamentally meaningless concept.
In that light, the new white paper looks more interesting, because (at least in my reading of it) it’s an attempt to reconcile the device’s real-world utility with the underlying science. It requires shedding some deeply ingrained assumptions about what power means. But even if you’re already a believer, grappling with the messy details of what’s under the hood of Stryd’s device might convince you that it’s even more useful than you thought.
Two Kinds of Power
Power is the rate at which you’re using energy. You can think of a runner as a machine that takes energy from food, and turns it into useful forces that propel you down the road. There’s a problem, though: no machine is perfect. You don’t get as much energy out as you put in. Cars, for example, are about 25 percent efficient: if you burn enough gas to get 100 joules of energy, only about 25 joules will go to spinning the wheels, and most of the other 75 joules will be emitted as heat.
Under normal circumstances, muscles are also about 25 percent efficient, but it varies widely depending on the specific circumstances. That means there’s a big difference between your input, which is known as metabolic power and reflects the food calories you’re burning, and your output, which is known as mechanical power and reflects how hard you’re slamming your foot into the road, how vigorously you’re swinging your arms, and so on.
I dug deep into this distinction and debate in an article back in 2018, and I took it for granted that we would all agree that runners and other endurance athletes are most interested in metabolic power, which is essentially a real-time estimate of how quickly you’re burning calories. Turns out not everyone agrees: “We don’t think most serious runners are all that interested in calories,” an engineer from Garmin, which has its own Running Power app, told me when I was reporting another article on running power.
I agree that runners don’t talk about calories much. But I think that’s mostly a question of terminology. If you go into a lab and use a bunch of sophisticated equipment to measure your VO2max, you’re basically measuring calories. You’re only interested in oxygen consumption because it’s a good proxy for how quickly you’re burning aerobic energy. And if you use that fancy lab data to identify a heart rate that will enable you to run at lactate threshold, you’re again using heart rate as a proxy for energy—i.e. calorie—consumption. And I would even argue that if you ditch all the technology and simply run by feel, trying to judge your pace so that you cover the prescribed distance as fast as possible, you’re relying on your perception of effort as a proxy for how quickly you’re burning calories.
Cycling vs. Running
No one gets tied into knots about this stuff in the cycling world. Power is power, and it’s considered the gold standard tool for effective pacing. The reason for this is that mechanical and metabolic power are almost perfectly correlated in cycling. If your power meter detects that you’re pressing 15 percent harder on the pedals, that means you’re burning calories 15 percent more quickly. The number on the display is mechanical power, but the reason people care is that it tells you what’s happening with your metabolic power.
Running, unfortunately, is totally different. Stryd’s white paper, which is written by in-house scientist Kristine Snyder with input from external scientific advisors Shalaya Kipp and Wouter Hoogkamer, identifies three reasons that mechanical and metabolic power don’t have a consistent relationship in running. One is that the motion of your limbs is far more variable than in cycling, which means muscle efficiency also varies more. The second is that each foot strike requires you to absorb forces rather than producing them, but you still spend metabolic energy cushioning these landings. And the third is that you store and then recycle energy in your spring-like tendons with each stride, boosting your mechanical power at no metabolic cost.
All of this would be irrelevant if you only ever ran on a smooth, level treadmill. The relationship between mechanical and metabolic power would be hard to calculate, but no one really cares about the exact relationship as long as the two powers move in sync. The problem is that once you step off the treadmill into the real world, the relationship changes. When you head uphill, for example, your stride gets less bouncy and as a result you get less free energy from your tendons.
Snyder, via email, gave me some illustrative numbers based on a recent journal article from a prominent biomechanics group in Italy. When you go from level ground to a 10 percent uphill gradient, your efficiency drops from roughly 60 percent to 50 percent. At a steeper gradient of 20 percent, efficiency drops even more to 40 percent. (Don’t get hung up on the exact numbers, which depend on which parts of the body you include in the calculation.)
In practice, this means that trying to maintain a consistent mechanical power while climbing hills would be a ridiculous approach to pacing. If you’re cruising along at 200 mechanical watts, an efficiency of 60 percent implies that you’re burning 333 metabolic watts. Once you’re climbing at 10 percent, maintaining the same 200 mechanical watts now takes 400 metabolic watts. You’re working about 20 percent harder even though the meter says your mechanical power output is constant! With that in mind, I don’t understand how any of the several companies that offer running power meters or apps can claim that mechanical power, on its own, is a useful metric.
What Runners Really Want
This is the reality that Stryd is formally acknowledging. Their device displays a reading that looks like mechanical power, calculated from a bundle of accelerometers, gyroscopes, a barometer, a wind probe, and other sensors packed into a foot pod. But the algorithm is explicitly designed to maintain a constant relationship between the number on the screen and your metabolic power. In the example above, if you maintained 200 watts on the Stryd device, you’d actually be producing 166 mechanical watts, which corresponds to 333 metabolic watts. Keeping the power constant on Stryd equates to keeping metabolic power constant and letting mechanical power change.
In the white paper, Snyder and her colleagues introduce a more subtle piece of terminology. What Stryd actually aims to provide, they explain, is a measure of instantaneous metabolic demand, rather than metabolic power.
For comparison, one of the key problems with heart rate is that it doesn’t respond instantly to changes in metabolic demand. When you start climbing a hill, your muscles begin consuming more energy immediately, but your heart rate drifts up more slowly as the body’s control systems respond to the change. This means that your muscles temporarily aren’t getting enough oxygen to meet their needs with aerobic energy, so they fill the gap with anaerobic energy. If you run up a hill while trying to keep your heart rate constant, you’ll sprint up the first section and only slow down once your lagging heart rate finally catches up to new demands.
Even in a fancy lab measuring your metabolic power with a VO2 machine, you’d encounter the same problem. Your oxygen uptake doesn’t respond instantly to changes like a steep hill. So Stryd aims to do better than the VO2 machine: it estimates how much metabolic energy your muscles are consuming in real time (metabolic demand) rather than how much energy your aerobic system is delivering, thus incorporating both aerobic and anaerobic energy contributions. In this sense, Stryd isn’t just mimicking what you could do in a lab; it’s doing something new and different—and, if you believe the data, better.
This opens up some intriguing possibilities, even beyond the ability to trust power for pacing when you go up a hill. Earlier this month, I wrote an article about the enduring controversy about what we mean by the term “threshold.” One of the conclusions was that the most relevant threshold definition for endurance athletes is something called critical power, which delineates the boundary between metabolically sustainable and unsustainable efforts. Critical power is a remarkably accurate predictor of performance in endurance races: top athletes, for example, tend to run marathons at about 96 percent of critical power.
You don’t necessarily need a power meter to work out your critical threshold. A study published earlier this year used Strava training data to estimate critical speed—that is, the speed that corresponds to critical power under normal conditions—for 25,000 runners. But “under normal conditions” is the catch. This approach works best if all the training data is collected on windless days on a level, smooth road, and your goal race is run under the same conditions. If those conditions aren’t met (and they never really are), then you’d prefer to use a metric that makes adjustments for things like wind, surface, and gradient. Stryd does that, and it automatically estimates a critical power for you based on your training data.
What’s on the Screen
This does leave one question unanswered. The number on the Stryd screen isn’t really mechanical power. It’s also not metabolic demand, though it’s proportional to it. So does it have any intrinsic meaning, other than as a proxy for metabolic demand? I went back and forth with Snyder on this multiple times, and each time she had to consult the Stryd team to avoid giving away proprietary information.
The closest I got to what I suspect is the real answer is this: “The scaling factor used is distinctly not arbitrary. It was chosen to allow consistency between power output values across activities.” I read that as a desire to have a power meter with a number that makes sense to cyclists, who already have strong intuition about what sort of power you might expect to sustain for, say, an hour. If you sold a device that simply displayed metabolic watts, it would create all sorts of cognitive dissonance for people who knew they could sustain 250 watts for an hour of cycling but were suddenly trying to sustain 1,000 watts for an hour of running.
I don’t necessarily think the Stryd team sat down and had that conversation when they were designing the device. As the 2015 article I mentioned at the top reveals, they were figuring things out as they went. The number on the screen probably does correspond to some particular portion of mechanical power, calculated in a particular way, under particular conditions. It takes some courage for the company to essentially say, “Forget about the number. The number is not important. It’s what it represents that matters.” But I think it’s the right call.
For more Sweat Science, join me on Twitter and Facebook, sign up for the email newsletter, and check out my book Endure: Mind, Body, and the Curiously Elastic Limits of Human Performance.