Three years of tinkering with performance tests: what we learned with Traiectum

An honest story about R&D: from a wildly over-ambitious idea to tests you can run with a phone.

It started in 2023, during a graduation internship at Mingle. The idea sounded simple: bring the performance tests that pro clubs use to grassroots football. No thousand-euro speed gates, no GPS vests, no force plates, just a phone, a few cones and a solid protocol. Back then we called them "minigames." Three years on, we've stood on plenty of pitches, and almost every idea we dreamed up behind a laptop has either been scrapped or completely rebuilt along the way.

That's exactly the story we want to share. Not because it all went smoothly, because it didn't, but because we learned an enormous amount. And because we're proud of where it ended up.

The lesson we kept relearning: the pitch is not your laptop

Our first version was, in hindsight, hopelessly over-engineered. We thought we could solve just about everything with AI. We even built a shot quality challenge: a test that used machine learning to measure where in the goal you shot and how hard. Beautiful, on paper.

Until a car parked on the street behind the goal with slightly-too-shiny rims. The model happily mistook those rims for a ball. An advertising board with round shapes on it? Also a ball. The challenge worked brilliantly, as long as conditions were perfect. And on a grassroots pitch on a Tuesday evening, conditions are rarely perfect.

Then we thought we were being clever with the sprint test. We'd make it easy to set up by using the lines already painted on the pitch as markers. No measuring tape needed, genius. At the very first club we brought it to, the pitch turned out to have... no lines.

It went like this time and again. What looks logical on a screen falls apart in practice. And the only reason we knew that is that we went out to the pitch from the very start. We never finished this behind a laptop and then went to test it; we were with teams from day one, sometimes with ideas that were still half-baked. And grassroots reality is ruthlessly honest: most teams have one hour of training and half a pitch to work with, and the test session simply replaced that evening's training. Everything had to fit inside that.

So we stripped things back. From six different tests to five, and eventually to three tests that still capture multiple physical qualities: a 30m sprint (acceleration and top speed), the Illinois agility test (change of direction) and the Yo-Yo Intermittent Recovery test (football-specific endurance). Fewer tests, far more usable. In that search alone we stood on a pitch with a team at least eight times, and many of those measurements were purely to learn from, not to keep.

Where Traiectum comes in

At some point we had a set of tests we believed in: this is the one. One thing had stayed the same from the very beginning: the tests have always been grounded in science. And that's exactly where our collaboration with Traiectumcomes in.

Traiectum is the sports-innovation ecosystem in Utrecht-East, a partnership between five elite sports clubs: FC Utrecht, SV Kampong, Hellas Utrecht, VV Utrecht and UZSC. Mingle is part of it as an innovation partner. Through Traiectum we could test our ideas with people who do this for a living: the performance team and sport scientists at FC Utrecht, and not least Eelco Veldhuijzen, former Head of Performance at FC Utrecht and now Sports & Performance Manager at Traiectum.

With them we checked not just whether we were measuring the right things, but something at least as important: are we measuring it accurately enough with a phone? Because a test can be scientifically sound and still be useless if the numbers are too far off. So we took our phones to Traiectum and put our measurements side by side with their "real" equipment: speed gates and force plates, the tools professional performance numbers come from.

The first time, we were still too far off, especially on acceleration and the jump. But by reviewing the videos, we worked out exactly what was wrong: not the phone, but our own protocol. We were leaning forward at the start, triggering the gate too early with a hand, and filming the jump at too low a frame rate to capture airtime properly. All fixable. We adjusted the protocol, came back, and the second time the sprint landed within about 3.5% of the professional equipment and the jump within about 2 to 3%. For a measurement taken with a phone and a few cones, that was a pleasant surprise. That validation gave us the confidence to take it onto the pitch.

The real work: evening after evening on the pitch

Going out to the pitch is something we kept doing, and it's perhaps the single thing we'd recommend most. Many evenings, at many clubs, across every level: from VV Kampong JO19 and a Vitesse senior side to girls' teams at VV Zwaluwen and ESA Rijkerswoerd, and DAW Schaijk's U13s. Setting out cones each time, getting the phones ready, and a team that didn't have a normal training that evening but our test session instead, and gave it everything.

But what those evenings really gave us wasn't the numbers. It was the conversations with the coaches. Every time, we asked: what are you running into? What's the reality for you? And that's where we learned the most. That a coach has no measuring tape but does have a pitch full of lines, or no lines at all. That there are only two volunteers to keep everything running. That an hour of training really is an hour, including setting up and the warm-up. Every evening we refined something, not behind a laptop, but based on what coaches told us.

Numbers only matter if you understand them

The more we tested, the clearer a second challenge became. A player who sprints 27 km/h or runs 1,200 meters on the Yo-Yo: is that good? For a coach without a sport-science background, a number like that means almost nothing on its own. And a test you can't interpret is a test you won't run a second time.

So we put at least as much time into how you present the results as into the tests themselves. We translated raw numbers into understandable categories, from "Poor" to "Excellent", into a team profile that shows at a glance where you're strong and where the development areas are, and into concrete training recommendations. Not "your acceleration is 20.8 km/h", but "as a team you score above average on speed, but acceleration needs attention, so here are a few drills". Otherwise it stays a number that disappears into a drawer after the session.

Why benchmark data matters so much

And that's where benchmark data comes in, something we ended up putting more and more time into. A score only means something once you can compare it: to peers, to players of the same gender, at the same level. Without that reference, every number is just noise.

The problem: for grassroots footballers, that reference barely exists in scientific terms. Almost all available benchmark data comes from the professional world or from controlled studies, which makes it poorly representative of a regular midweek amateur team. And for women and girls it's even thinner: there's structurally little data, even as women's football grows fast. A 15-year-old player deserves a fair comparison with her own peers, not with a male professional.

That's why we're now building that reference ourselves, evening after evening, team after team. By age group and gender, our dataset is growing, and the more teams take part, the sharper and fairer the benchmarks become.

The honest truth about R&D in a startup

We'd love to tell you that we then also ran the big test days with more than twenty teams at once, exactly as we'd planned. But that's not what happened. It simply turned out to be logistically hard to schedule that at scale with clubs. Then came the summer break, and after that the day-to-day reality at Mingle got in the way. As it often does.

And that's perhaps the most honest lesson of this whole journey: this is just how R&D works inside a startup. You kick off a project, it runs hot for a while, then it slips into the background, and at some point you pick it back up. Not every plan ends where you imagined it. But you pick up a mountain of insight along the way, and that insight is now baked into something concrete that sticks around.

And now: curious?

Everything we've learned over the past three years, we're gradually collecting in a performance testing guide for grassroots coaches: the tests themselves, run with basic equipment and a phone, plus the protocol lessons we learned the hard way (the speed gate at 1m, filming at the right frame rate, how to recognize a valid jump, how to process the videos and read the results). It's very much a work in progress, just like the whole project, and we keep refining it as we learn more.

Fancy trying the tests yourself, or just want to know a bit more about how they work? You're very welcome to take a look.
👉 Check out the Performance Testing Guide

And if you want to think more broadly about player development: in our ebook From Grassroots to Greatness, nine experts, including Eelco himself, share their insights on how to make players better, even with limited time and resources.
👉 Download the ebook