Amateur football’s problem is not talent identification. It is that most of it was never written down
Below the top two tiers of most European federations, "how much football happened" and "how much football left a record" are very different quantities — and a career can turn on which side of that gap a player falls.
David Avramovic 3 min
Scouting is usually described as a search problem: there are good players nobody has found, and the job is to find them. That framing is half right and the missing half is the expensive one. A great many of those players have already been watched, repeatedly, by people who thought they were good. What did not happen is that anybody wrote it down in a form that survived the conversation.
A coach’s opinion lives in a coach’s head. A parent’s video lives on a phone. A district league’s results live in a spreadsheet somebody maintains as a favour. None of that is a record a sporting director two divisions up can act on, and none of it survives the coach changing club.
What we can measure about this, and what we cannot
PlayMaker cannot tell you how much football is played in Austria. Nobody can. What it can tell you is how much football has left a record on this platform, per competition, per registered player — and how much of that record anybody other than the player stands behind.
That is a narrower claim than it sounds and it is deliberately narrow. It is a fact about PlayMaker’s own coverage, not a fact about Austrian football, and the two get conflated constantly in this market. A platform reporting that "Regionalliga players average X minutes" is usually reporting how many Regionalliga players happen to use it. The figure worth publishing is minutes per registered player, which is at least a property of the cohort rather than a property of our marketing.
Why the withheld count is printed
A table that quietly drops its small rows tells the reader it covers everything. That is the same lie as a search result capped at fifty and presented as the whole list, and it is a lie the reader has no way to detect. So the suppression is stated: how many rows were withheld, and how many players they covered.
The floor exists because a cohort small enough to identify somebody is not an aggregate, it is a disclosure wearing an average’s clothes. "The one U19 league in the state with two players, both signed" names two children to anybody who follows that league. Five players is the usual floor for this class of small-area statistic and it is a single constant, because a per-table floor is a floor somebody will lower for one harmless-looking table.
The outcome data is the part that gets interesting
Every scouting platform can tell you who exists. Almost none of them can tell you what happened next, because they have no relationship with either side after the introduction. PlayMaker records contacted, replied, trialled, signed — or passed, with a reason — and reports it per competition.
Conversion is a better measure of whether a league produces professionals than any average rating, because it counts what actually happened rather than what a model predicted. It is also the dataset that makes a model worth building later, which is the honest order of operations: outcomes first, then a model trained on them. A platform that leads with "AI matching" and has no outcome data has an unexplainable model and nothing to check it against.
None of this is a ranking of children
It never will be. The public surface here is aggregates about competitions and clubs, plus a page for adults who separately asked for one. There is no directory of players, no list endpoint, no "similar players" rail on a public page — each of which would quietly turn this into the thing it was built not to be. The players on this platform are mostly minors who consented to being seen by verified scouts, and no amount of search traffic stretches that consent to the open web.