The Runners Who Get Faster Are Building, Not Grinding
Terra found about half of runners never improve their aerobic fitness. I checked our own 122 committed runners and asked what the half who did get faster were doing differently. It wasn't more miles or harder sessions. It was progressively building their training. And the runners on an athletedata plan were the ones building, and getting faster, while the runners mostly just tracking stayed flat.
The Question Worth Asking
Terra Research ran a study this month with a blunt title: most runners don't actually get faster. They started from 856,000 running activities, kept the 641 runners with enough history to track, and found only about half had improved their aerobic fitness over time. The other half ran for months and stayed where they started.
You can read that as bad news, or you can ask the more useful question. If half of runners get faster and half don't, what is the half who improve doing that the other half isn't? Because whatever that is, it's the thing worth copying. I had the same kind of data in our own tables, so I went looking for it.
First, an Honest Correction for the Season
Terra tracked each runner for a median of three and a half years. We are younger than that, so most of our runners have six to twelve months of history, and a lot of it ends right now, in June. That matters, because running at the same heart rate gets slower in heat. We measured it directly two weeks ago: our runners give back real pace at the same effort once it warms up, and it is the weather, not lost fitness. You can read that one here.
So before I judged anyone's trend, I removed the seasonal pattern everyone shares. For each month I worked out how much faster or slower the whole group ran than their own normal, and subtracted that swing. The result is what you'd expect: our runners are most efficient in cool April and lose the most in January and June. With that gone, a hot summer no longer poses as a wasted year. I re-ran everything on weather-adjusted pace as a second check, and it agreed.
How I Measured Getting Faster
The cleanest sign that an aerobic engine is growing, short of a lab, is speed at a given heart rate. Coaches call it efficiency factor. Faster this month than last for the same heartbeats means fitter. It is the measure I used in the easy-running post and the consistency post, and it is the same idea Terra used to define getting faster.
I kept runners with at least 20 genuinely aerobic runs over at least six months. That left 122 runners and 7,614 runs. For each one I compared their earliest runs to their latest, after the seasonal correction, and asked whether they ended faster at the same heart rate. These are committed runners, not weekend joggers. The median one logs about 140 km a month across a dozen runs.
The Lever Is Building, Not Bigger or Harder
Here is the finding, and it surprised me.
The runners who got faster were not the ones running the most. Split our 122 by monthly volume and the highest-mileage fifth improved about as often as the lowest. Among people who already train seriously, how much they ran barely told you who improved. The same was true for intensity. The runners leaning hardest on hard sessions were, if anything, slightly less likely to be getting faster, which lines up with Terra and with a lot of coaching sense. You don't climb out of a plateau by hammering.
What actually separated the two groups was whether a runner was building. I split them by their volume trend, comparing each runner's later months to their earlier months. The runners who grew their training across the window were the ones getting faster, and the runners who held flat were the ones stuck. Lined up from "let it slide" to "built the most," the share who got faster climbs from 33% to 69%. The runners who built their volume by more than 15% had a median that moved the right way. The ones who held roughly steady mostly got a touch slower, even though they were still putting in real miles.
That is the whole thing in one line. Holding a steady, already-decent training load keeps the fitness that load built. It does not keep building it.
Part of this is regression to the mean, and I want to be straight about that. Runners who started further from their ceiling had more room, and our slower starters did improve more. I controlled for where each runner began and the building effect held, so it isn't only headroom. Runners who were actively progressing got faster more often than runners parked at a steady number.
Why Building Is the Hard Part to Get Right
None of this is exotic. It is progressive overload, the oldest idea in training. Your body adapts to a workload and then stops treating it as a reason to change. Hold the routine and the stimulus that built you becomes the one that maintains you. Coaches watch strength plateau within weeks on a fixed program and endurance plateau over months on a fixed weekly load, which is about the window we caught these runners on.
The catch is that building well is genuinely hard to manage alone. Add too little and you maintain. Add too much, too fast, and you buy the thing that erases progress. When I looked at consistency, the single worst event for efficiency was a long gap, the two-week hole a cold or a niggle punches in your block. Ramp carelessly, get hurt, and the forced break costs more than the miles you gained. Sit too long fully off and fitness does fade too, which I dug into here. So the right amount of building lives in a fairly narrow band, and finding it week to week, around your sleep and your life and your last hard session, is exactly the part most self-coached runners get wrong.
The Runners on a Plan Were the Ones Building
This is the part I cared about most, because it's the difference between a nice chart and something you can act on.
Our runners split cleanly by whether they were using athletedata's coaching. Of the 122, the ones following a structured plan (10 or more planned sessions) were building 86% of the time, versus 76% for the runners mostly just tracking their data. More striking, the plan group built about twice as much over the window. And it showed up where it counts: the median runner on a plan got faster, while the median runner who was mostly just tracking got slightly slower. By the headcount, 52% of the plan group improved against 46% of the rest. The runners who set a goal in the app told the same story, building far more and improving more often than the runners who didn't.
I'll be honest about what this is and isn't. Runners who commit to a plan are more motivated to begin with, so this is an observed pattern, not a controlled trial, and some of the gap is who chooses to follow a plan in the first place. But the mechanism is the same one the whole dataset keeps pointing at. A plan exists to do exactly the thing that separates the runners who get faster: add load progressively, at a rate your body can absorb, and hold the line when you need to recover instead of when a calendar says so. The runners doing that were the ones improving. The ones improvising were the ones drifting.
What I'd Do
If you feel stuck, the data points away from the two moves runners usually reach for. A harder interval session didn't separate the people who improved, and gritting out the same week you've run for a year went right along with staying flat. What moved people was progressively growing the load they could absorb. Add a little, let it settle for a few weeks, add a little more, and protect the streak so you never hand the gains back to a long break.
If you're happy where you are, holding steady is a fine choice. It mostly keeps what you have. Just don't expect the line to keep climbing on its own, because for half of a very committed group, it didn't, and the ones it did climb for were building on purpose.
Why I Care About This at athletedata
A crowd average can tell you "run more and be consistent," and it's not wrong, but it can't see whether you personally are adding to your engine this month or quietly maintaining it. Your own efficiency curve can. That's the job athletedata does. It reads your training and recovery, watches that curve, and paces the build for you, nudging the load up when your body is ready and easing off before a ramp turns into a forced two weeks down. The runners in our own data who let it do that job were the ones getting faster. That is the coaching that comes from your data instead of from everyone else's.
athletedata is an AI running and triathlon coach that reads your Garmin, Strava, Oura, and WHOOP data and builds a plan that adapts to your training and recovery. Try it free.
