Cadence Is the One Form Metric Worth Changing First

Somewhere between 2010 and now, “180 steps per minute” stopped being an observation and became a commandment. It shows up in Reddit threads, in the tooltips of coaching apps, in the mouths of people at parkrun who have never met you. And if you are following a Runna block or a Garmin Coach plan and your watch keeps telling you that your cadence is 162, you have probably wondered whether you are doing it wrong.

You are almost certainly not doing it wrong. But there is a version of the cadence advice that survives scrutiny, and it happens to be the single most self-coachable form change available to a recreational runner. That version is not “get to 180.” It is “go about 5% higher than whatever you currently run, at easy pace, for a few weeks, and see what your knees think.”

Where 180 actually came from

Jack Daniels, the exercise physiologist, sat in the stands at the 1984 Los Angeles Olympics and counted the steps of distance runners in finals. He found that nearly all of them took 180 or more steps per minute. That is the entire origin story. It was a count of elite athletes, most of them small and light, racing at somewhere between 4:30 and 5:30 per mile in an Olympic final.

Nobody in that sample was doing a Tuesday recovery jog at 11:30/mi. Nobody in that sample was 5’11” and 185 pounds. Daniels himself has spent years pointing out that the number was descriptive, not prescriptive, and that cadence rises with speed for basically everyone. The 180 figure escaped into the wild anyway, because round numbers are sticky and “it depends on your height, mass, leg length and current speed” does not fit on a shoe box.

Here is what the data on actual recreational runners looks like. Studies of non-elite runners at comfortable paces have found mean cadences clustering somewhere in the 155 to 170 range, with individual variation of 20+ spm in either direction at the same speed. Leg length explains a chunk of it. Height explains a chunk of it. Pace explains the largest chunk of all: your own cadence at 6:30/mi is not the same as your cadence at 10:00/mi, and it shouldn’t be.

So the tall runner plodding along at 158 on a recovery run is not broken. They are a tall runner on a recovery run.

Why cadence is still worth touching

Here is where I stop debunking and start advocating, because the useful finding got buried under the myth.

When researchers take a runner’s own habitual cadence and nudge it up by 5 to 10%, holding speed constant, a consistent cluster of things happens:

• Step length shortens, so the foot lands closer to under the body rather than out in front.
• Braking force drops. The horizontal impulse that decelerates you on every landing gets smaller.
• Vertical oscillation drops. You bounce less, which means less mass to catch on landing.
• Peak knee flexion angle and knee joint loading drop. The commonly cited figure for a 10% increase is around a 20% reduction in energy absorbed at the knee per stride. At 5% it’s smaller but still measurable.
• Hip adduction and contralateral pelvic drop reduce, which matters for anyone with a history of IT band trouble.

The studies behind this (Heiderscheit and colleagues at Wisconsin, Lenhart, Willy, and a good number of replications since) are lab studies on treadmills with modest sample sizes. They are not twenty-year injury-outcome RCTs. What they show is a biomechanical mechanism, reasonably consistently, across labs. That’s a weaker claim than “this prevents injuries,” and I want to be honest about it. But it’s a considerably stronger evidence base than almost anything else you’ll be told about your form: forefoot striking, arm carriage, “run tall,” pulling your heel to your glute, any of it.

And crucially, cadence is the only form variable that your watch already measures, reports per-interval, and can beep at you about in real time. That combination is rare. You can self-coach it without a treadmill, a physio, or a slow-motion camera.

The 5% rule, worked

Stop trying to hit a target and start working from your own baseline. Here’s the procedure.

Step 1: Find your real easy-run cadence. Not your race cadence, not your average across all runs (which blends paces and pollutes the number). Pull up three or four recent easy runs in Strava or Garmin Connect, ignore the first half mile while you’re settling in, and read the average cadence over a steady middle section.

Say you get this:

Easy run, 5.2 mi @ 10:14/mi
  Avg cadence (full run):        164 spm
  Avg cadence (mi 2-4, steady):  162 spm
  Avg vertical oscillation:      9.8 cm
  Avg ground contact time:       282 ms
  Avg stride length:             1.21 m

Your working baseline is 162, not 164.

Step 2: Multiply by 1.05. 162 × 1.05 = 170.1. Round to 170. That’s your target for easy runs only. Note that this is nowhere near 180, and that is completely fine. If your baseline were 172, your target would be 181, and you’d have sailed past the magic number without ever aiming at it.

Step 3: Hold pace constant. This is the part people get wrong. If you increase cadence and increase speed at the same time, you have changed nothing about your mechanics; you’ve just run faster. The whole point is shorter steps at the same pace. Your stride length should drop from 1.21 m to roughly 1.15 m, and your GCT and vertical oscillation should come down a little too. If your pace jumped 30 seconds per mile, you ran a different workout.

Step 4: Set the watch up to do the nagging. On a Garmin, add Cadence to a data field on your easy-run screen, then build a cadence alert: Menu → Activity & Apps → Run → Alerts → Add New → Cadence → set a low threshold of 168. You’ll get a buzz when you drift under. On a Coros, it’s Run → Settings → Alerts → Cadence. Apple Watch users can get this through WorkOutDoors, which does custom cadence alerts far better than the native Workout app.

Step 5: Use a metronome for the first three or four sessions. Set it to 170 (or 85 and land on every click, which some people find easier to hear). The Metronome app that ships inside Garmin Connect IQ works, as does any free phone metronome, as do Spotify’s running-BPM playlists if you’d rather have music. Four to six weeks in, most runners find the new cadence has become automatic on easy days and has quietly pulled their tempo and threshold cadence up along with it.

Step 6: Change one thing at a time. Don’t do this in the same block where you’re switching to carbon plates, adding a second quality session, or trying to land midfoot. If something starts hurting, you want to know what caused it.

What this looks like against an AI plan

The reason this matters for anyone running an algorithmically generated block: cadence work is the rare thing you can layer on top of a Runna or Garmin Coach plan without touching the plan itself. It costs no extra load. It changes no session. It replaces nothing.

So a sensible four-week overlay on a marathon build looks like this:

After four weeks, compare like with like. Find an easy run from week 0 and an easy run from week 5 at the same pace on the same route, and put the numbers side by side:

                        Week 0        Week 5
Pace                   10:14/mi      10:11/mi
Cadence                 162 spm       171 spm
Stride length           1.21 m        1.14 m
Vertical oscillation    9.8 cm        8.9 cm
Ground contact          282 ms        271 ms
Avg HR                  142 bpm       143 bpm

That’s the shape of a successful change: cadence up roughly 5%, stride length down proportionally, oscillation and contact time down modestly, heart rate flat or near enough. If your heart rate has gone up 8 beats at the same pace, you are overstriding in the other direction (too short, too choppy) and should back the target down to 3%.

One honest caveat: your watch’s cadence number is good, but vertical oscillation and ground contact time are estimates from a wrist or chest accelerometer, and their absolute accuracy is mediocre. Treat them as trend lines on your own data, not as measurements you’d compare against a friend’s.

If you want to see what’s actually happening to your foot strike and knee angle rather than inferring it from wrist-derived metrics, that’s a job for video. Our guide to AI gait and form analysis from phone video walks through how to film yourself properly and what the analysis tools can and can’t tell you, which pairs well with a cadence block because it gives you before-and-after footage rather than just before-and-after numbers.

Who should skip this

Runners already at 175+ on easy runs at conversational pace. You’re fine. Adding 5% puts you at 184 and into territory where the metabolic cost of swinging your legs starts to outweigh the mechanical benefit.

Anyone in the last six weeks of a marathon build. Introducing a novel motor pattern while you’re doing your highest-volume weeks is asking for a calf strain. Park it for the off-season.

Runners with a current Achilles or calf problem. Increasing cadence tends to shift load toward the ankle and calf as it shifts load away from the knee. If your complaint is patellofemoral pain, ITB syndrome, or tibial stress, cadence is working in your favour. If it’s Achilles tendinopathy, it isn’t, and you may want the opposite adjustment.

Anyone who has changed three things this month already. Pick one.

The bit nobody tells you

The reason this intervention works better than every other form cue is not really biomechanical. It’s that cadence is countable. “Run tall” is unfalsifiable. “Land under your hips” is something you cannot perceive accurately while doing it. But 170 either is or isn’t on your wrist, and you’ll know within thirty seconds of starting a run whether you’re holding it.

Motor learning research is fairly clear that external focus cues (a metronome click, a number on a screen) produce faster and more durable change than internal cues (think about your knee). Cadence is the only running form variable that comes with a built-in external cue you already own. That’s why it’s first. Not because 180 is holy, but because 170 is achievable, measurable, and specific to you.

Go and look up what your last three easy runs actually averaged. The number you find is the only starting point that matters, and it’s almost certainly not the one in the tooltip.