From one school to a district

A pilot that works in one school proves less than it appears to. The first school has the founders walking its corridors, a teacher who answers messages at night, and a group of friends who want the project to succeed. None of those advantages travels. The interesting question is not whether FitForm works at the school that built it; it is what has to be true for it to work at a school where nobody on the team has ever been.

The first thing that breaks is explanation. In the pilot, a new class learns the system from a person. At the tenth school there is no person, so everything that person said must exist as something else: a one-page guide on the wall beside the station, a code that opens a two-minute video, and an interface that makes the next action obvious to a student who has read nothing. The team discovered that the single most effective change was not a feature but a sentence — telling students exactly where to stand, before the camera starts.

The second thing that breaks is the station itself. One tablet in a hall serves perhaps forty sessions in a break if everything is smooth, which it is not. Scanning a card takes four seconds, getting into position takes ten, and the queue decides whether the system feels fast or hopeless. The fix is boring logistics: two stations instead of one, a line marked on the floor at the right distance, a charger that stays plugged in, and a rule that the station belongs to one class at a time in a published timetable. Software cannot solve a queue.

The third is the governance of data. In one school, a teacher knows every student and nobody worries about a name on a board. Across a district, the same board is a list of children's names and schools visible to anyone with the link. The project's answer has been to keep the cheap, strict version from the beginning: no video leaves the device, the public board shows a first name and a class rather than a full name, an account can be deleted by a teacher in one click, and nothing is shared with any third party. Such a policy is far easier to keep than to introduce later, because every exception that exists becomes a habit someone defends.

The fourth is the thing nobody plans for: who owns it next year. The three founders will leave grade nine, and a project that depends on them will stop with them. Making it survivable is unglamorous work. The deployment steps live in a file in the repository, so a new student can put the server back after a mistake. The tests run in one command, so a change that breaks counting is caught before a competition, not during one. The club writes down how a leader is chosen, and each new member is given one small, real responsibility in their first month, which is how a committee becomes a team.

What does success look like at district scale? Not the largest number the team can print. A better set of targets is modest and checkable: three schools where at least half of one grade finishes a first session, a median of three sessions a week for active students in week twelve, one teacher per school who can restart a station without help, and nobody anywhere who can show that a published number was wrong. If those four things hold, growth is mostly a matter of time and photocopying. If they do not, a hundred schools would only multiply the problem — which is the real lesson of every pilot that looked wonderful and then quietly stopped.