How football clubs should evaluate new performance technology

Football clubs are not short of offers. A new wearable, a new dashboard, a new model that promises to tell staff something they did not know about their players. Some of these tools are excellent. Many are not. Most sit somewhere in between, useful for one question and misleading for others.

We are a vendor too. BOUND is building a smart insole, and its V1 prototype boards are currently undergoing bench validation. So this is not a neutral piece, and it should not pretend to be. What follows is the checklist we think any club should apply to any performance technology, including ours. Where the evidence behind each question comes from published work, we say so. Where it is our opinion, we say that too.

Start with the decision, not the device

The most useful framework we have found for this comes from Windt and colleagues, who propose that a team ask four questions in order before a purchase: would the information help, can you trust it, can you integrate and analyse it, and can you implement it in practice. Only if all four are answered yes does a cost-benefit judgement make sense ¹.

The order matters. The first question is not “what does this device measure?” but “what decision would we make differently with this information?” A physio deciding when a player returns to full training needs different information from a head of performance planning a congested month. If nobody can name the decision, the most accurate sensor in the world becomes another number on a screen.

It is also worth asking whether an existing data stream already answers the question. Windt and colleagues make this point directly ¹. It is an uncomfortable one for vendors, and a necessary one for clubs.

Ask what claim is being made, and test that claim

Vendors tend to talk about devices. The more useful conversation is about claims. A recurring point in the validation literature is that validity belongs to the inferences drawn from data for a particular purpose, not to a piece of hardware ¹. The clinical V3 framework makes the same separation more formally: whether a sensor records what it should (verification), whether an algorithm turns those recordings into an accurate metric (analytical validation), and whether that metric means something useful in a defined context (clinical validation) ². Football is not a clinical trial, but the distinction carries over well.

In practice this means asking a vendor three separate things. Has the sensor been tested? Has the metric been compared against an accepted reference, in movements that look like football rather than a treadmill? And has anyone shown that the metric helps with the decision you named in the first step?

Düking and colleagues give a practical bar for the second question. A device’s typical error should be compared with the smallest change that would actually matter, and agreement with a criterion should be reported with appropriate statistics rather than a correlation alone ³. Hopkins and colleagues describe widely used thresholds for what counts as a small but meaningful difference ⁴. If a tool’s noise is larger than the change you care about, it cannot tell you about that change, however polished the dashboard.

One governing-body detail is worth knowing here. Wearable tracking systems have been permitted in matches since a 2015 change to the Laws of the Game, with the final decision left to each competition ⁵. FIFA runs a testing programme for electronic performance and tracking systems, with a safety tier and a performance tier for positional tracking. FIFA states that approval of the hardware is not an endorsement of the quality of the information the device produces ⁶. The performance standard covers positional tracking. It does not cover in-shoe force or pressure metrics, so BOUND, like any insole maker, cannot point to a FIFA quality mark for what it measures. It will need its own published validation.

Demand the raw data and the version history

This is the point where we are most opinionated. We think clubs should make raw data access, and disclosure of firmware and software versions, a condition of adoption.

The published basis is modest but clear. Windt and colleagues raise whether a club can obtain raw data or only processed summaries ¹, and Düking and colleagues recommend reporting firmware and software versions, sampling and filtering, and raw-data access when evaluating wearable data ³. Our view goes further: when a vendor changes an algorithm, a club with only processed numbers cannot tell whether its players changed or the software did. A club with the raw data and the version history can.

Be sceptical of injury claims

If a vendor promises to predict or prevent injuries, ask for the evidence, and read it carefully. Impellizzeri and colleagues argue that research linking training load to injury, including popular load ratios, is not suitable for deciding how to reduce injury risk, and that load data alone cannot show whether a change is raising or lowering that risk ⁷. That argument is contested by some researchers. But the burden of proof sits with whoever makes the claim. We are not aware of evidence that adopting any wearable reduces injuries in football, and BOUND makes no injury claim.

Plan for the people, not just the data

A technically sound tool can still fail. Windt and colleagues note that implementation problems, including staff workload, athlete burden and a lack of buy-in, can sink an otherwise good technology ¹. The athlete side has some direct evidence. In a small qualitative study of elite sprinters, the main reason athletes disengaged from monitoring was a lack of feedback from the staff collecting their data ⁸. That study is small and outside football, so we treat it as a warning rather than a rule.

It is also worth remembering that the newest tool is not automatically the most responsive one. A systematic review found that athletes’ own self-reported measures responded to training load more consistently than several commonly used objective measures ⁹. Simple tools deserve a place in the comparison.

Agree on who owns the data before day one

Players increasingly expect rights over data collected about them. FIFPRO’s Charter of Player Data Rights, developed with FIFA, sets out expectations including access, correction, deletion, portability and the ability to withdraw consent ¹⁰. It is a players’ union position rather than law, but it describes where the conversation is heading. Clubs should know, before a pilot starts, where data is stored, who can see it, how long it is kept and how a player can get it removed. With youth players, consent and data handling need to be settled with families and the club before any recording.

The short version

Before adopting any performance technology, a club should be able to answer five questions:

1. Which decision will this change? Name it, and check no existing tool already answers it.
2. What exactly is being claimed, and against what reference was it tested? In football-like movements, with error compared to the smallest change that matters.
3. Can we get the raw data and the version history? If not, ask why.
4. Who will use it, and will athletes get feedback? If nobody has time to act on it, it will not last.
5. Who owns the data, and how do players exercise their rights? Settled in writing, before the first session.

These are the questions we expect clubs to ask us. They are the right ones to ask everyone.