In May 2024, I wrote a newsletter called Running With Power. By then, I had owned a Stryd for around eight years.

I bought one of the first-generation models in 2016, largely because I came to running with a cyclist’s curiosity about power. Power had transformed cycling training, so surely there was something in it for runners too?

The question I was asking was simple: is power a better way to measure running intensity?

Two years later, I think I may have been asking too narrow a question.

A few weeks ago, I had an exchange with David Roche. David and Megan Roche host the excellent Some Work, All Play podcast and coach some of the best runners in the world.

I asked David about running power, and his criticism was sharp. Running power meters, he argued, can create an illusion of precision.

Your watch says 300 watts. Not 295-ish, not somewhere around threshold, not comfortably hard. Exactly 300.

Running power is not measured through a drivetrain in the way cycling power is. It is modelled from speed, acceleration, gradient, air resistance and a set of proprietary assumptions. Perhaps that 300 could just as easily be 290 or 310.

That does not make the number useless. On hills, in wind or over variable terrain, power may be far more informative than pace. It can also give a developing runner an external reference when their feel is not yet well calibrated.

But if pace, heart rate and perceived effort already tell an experienced runner roughly the same thing, what has the power number really added?

The more I have thought about it, the more I think David has a point. Perhaps a very good one.

He later described coaching as:

I loved that, and I still do. But one question has been nagging at me since.

What is a developing runner supposed to feel?

Oddly, I had already written part of the answer in that 2024 newsletter. I described RPE, rating of perceived exertion, as a learned skill.

A good runner is constantly triangulating. Pace says one thing, heart rate says another and perceived effort says something else. Perhaps they slept badly. Perhaps it is hot. Perhaps yesterday’s long run is still sitting in the legs.

The numbers do not quite agree, but the experienced runner adjusts. They become comfortable in the chaos.

Most of those signals, though, are trying to answer one question: how hard am I running?

Easy, steady, threshold, too hard, back off.

A developing runner has a much bigger triangulation problem because they are also trying to answer: how am I running?

They are trying to interpret rhythm, ground contact, stiffness, posture, tension, projection, stride length, shoes, terrain and pain, often all at once.

We tell developing runners to “run by feel” before they have developed much of a sensory vocabulary. Their feel may simply be: this is hard. Or: my Achilles hurts.

This matters to me because of how I came to running.

In my 30s, I would have struggled to run 10 kilometres in under an hour. When I started swimming, I was immediately struck by how seriously the sport treated technique.

Nobody looked at a poor swimmer gasping after 25 metres and thought: they just need a bigger aerobic engine.

We talked about balance, body position, catch, timing, breathing and movement. I remember thinking there must be something similar in running.

So I started experimenting. I moved away from the highly cushioned shoes I was wearing and began paying much closer attention to how I moved. Within months, the difference was remarkable.

My biggest early leap in performance did not come from finding a magic training zone. It came from changing how I ran.

Form before fitness.

That experience has shaped almost everything I believe about coaching.

I see this constantly in swimming.

An adult learner with sinking legs does not think: my centre of mass and centre of buoyancy are poorly aligned. They simply feel tired. Swimming feels difficult.

Then we give them a balance drill. For a few seconds, they experience a different position in the water, and we say: remember that feeling.

The drill has not magically made them fitter. It has expanded their sensory vocabulary. They now have two movement solutions to compare.

Running may be similar.

You cannot feel a movement option you have never experienced. More importantly, your feel is calibrated to the runner you already are.

Recently, I started a small experiment with a group of athletes using the Norwegian Singles approach.

The structure is wonderfully repetitive:

Tuesday: 3 × 10 minutes

Thursday: 5 × 6 minutes

Saturday: 10 × 3 minutes

All controlled sub-threshold work.

At first, my interest was mainly physiological. Could busy amateur athletes improve by protecting three repeatable sub-threshold sessions each week? Could we control the intensity well enough to accumulate quality work without constantly tipping over into threshold or VO2max training?

But the repetition created something I had not really considered: information.

Think about a more varied running programme. One week you do 12 × 400 metres. The next week hills, then 5 × 1 kilometre, a progression run, a fartlek or a broken tempo.

All potentially excellent sessions.

But now try to analyse how the athlete moves. Was cadence lower because the reps were longer? Was ground contact time higher because the pace was slower? Was stride length different because of the hill? Was it the shoes, the surface, the recovery or fatigue?

We change six variables and then stare at the data, hoping it will reveal a pattern.

The Norwegian Singles structure was different. The same sessions, in the same intensity domain, repeated week after week.

I had accidentally turned the workout into a laboratory.

One of the athletes in the experiment is Daniel.

Daniel is 195 cm tall and has had some Achilles problems. Last Saturday, he completed his 10 × 3-minute session.

I noticed on Strava that his cadence looked low. Daniel also owns a Stryd, so I opened the data.

His cadence across the ten intervals was:

155, 155, 155, 156, 156, 156, 156, 157, 157, 157.

His power sat almost metronomically between 394 and 408 watts. Stride length was around 1.20 to 1.24 metres, while ground contact time sat between 307 and 318 milliseconds.

My first thought was not that 156 was bad. There is no magic cadence, and Daniel is also 195 cm tall. Comparing him blindly with some mythical 180-steps-per-minute ideal would be misleading.

What struck me was something else.

Daniel had reproduced almost exactly the same movement solution ten times.

I had completed exactly the same workout.

My cadence across the ten intervals was:

176, 174, 173, 176, 174, 176, 176, 177, 177, 177.

My stride length was around 1.22 to 1.30 metres, and ground contact time was around 240 to 260 milliseconds.

I am 179 cm tall and was running around 44 seconds per kilometre faster than Daniel. Yet our absolute stride lengths were surprisingly similar.

We were doing the same workout but solving the movement problem very differently.

This does not prove that my mechanics are right and Daniel’s are wrong. Speed explains some of the difference. Height matters, anatomy matters and running history matters.

But suddenly I had a much better question: what happens if Daniel increases his cadence by around 5% while holding the rest of the workout broadly constant?

This brought me back to my conversation with David Roche.

Perhaps he is right about running power. Perhaps 400 watts is not really 400 watts. Perhaps the number carries more precision than the underlying model deserves.

But looking at Daniel’s data, I realised something.

For perhaps the first time in ten years of owning a Stryd, I was not particularly interested in the big number labelled POWER. I was interested in how Daniel moved: cadence, stride length, ground contact time, leg spring stiffness, vertical movement and, most importantly, the extraordinary repeatability of the pattern.

Perhaps I had spent ten years concentrating too heavily on the headline number.

Power may still be extremely useful. It can help regulate effort on hills, expose changes that pace hides and provide structure to runners whose internal sense of intensity is still developing.

But maybe Stryd’s most interesting value is not only telling a runner how hard they are running. Perhaps it can also help a developing runner begin to understand how they create speed.

Daniel’s next session is 3 × 10 minutes.

The experiment is simple. The first ten minutes are natural. The second ten minutes are at 164 to 165 steps per minute. The final ten minutes return to his natural cadence.

The power target stays the same.

The cue is:

We are not trying to prove that 165 is better than 156. I want Daniel to experience another movement solution.

What happens at the ankle and hip? Does he feel quicker off the ground? Does the stride feel constrained or smoother? What happens to stride length, ground contact time and stiffness? And what does his Achilles say the following morning?

The numbers are not replacing feel. They are helping us create and label a new feeling.

Daniel recently described a new pair of very bouncy shoes to me. He talked about needing to push off, feeling the shoe absorb the force and then being launched forward. He felt more in his hip, less stable around the ankle and thought the shoe somehow gave him a longer stride.

That is an athlete trying to triangulate movement. The data gives us another point in that triangulation.

We often say consistency is king because consistency creates adaptation.

That is true, but consistency has another superpower.

Consistency creates information.

If every workout is different, every data point needs an explanation. Repeat the same controlled session and patterns begin to emerge.

Same workout. Same demand. Change one thing. Feel it. Measure it. Repeat.

Eventually, perhaps, you no longer need the measurement.

This has also made me rethink our Norwegian Singles experiment.

We are building a more comprehensive app to help us analyse the training data. Until last week, I assumed the interesting data would mostly be intensity data: power, heart rate, pace and drift. The things that help us understand whether the athlete is controlling the work.

Now I am wondering whether we should be looking at something else too: movement.

If the same athlete repeats 3 × 10, 5 × 6 and 10 × 3, week after week, at a controlled metabolic demand, perhaps cadence, stride length and ground contact time start to tell their own story.

These are movement signals rather than a full biomechanical analysis, but repeated under similar conditions they may still become useful.

Not because there is a perfect cadence. Not because 250 milliseconds of ground contact is good and 300 is bad. That would simply create a new illusion of precision.

But perhaps change against your own repeated movement fingerprint is information.

I genuinely do not know yet, which is exactly why it is an interesting experiment.

We may have started the NS Project to study training consistency and accidentally built a better way to observe movement.

David Roche described coaching as “guiding feel and being comfortable in the chaos.”

I love that. But the more I coach adult swimmers and developing runners, the more I think there is a step before it.

First, we have to teach athletes what is available to feel.

A good runner may be triangulating pace, heart rate and perceived effort almost subconsciously. A learner is triangulating something much bigger.

Not just metabolic effort, but movement itself.

Maybe the most valuable workout is not the one that gives you the biggest stimulus.

Maybe it is the one you know well enough to notice when something changes.