We built a membership system for a gym this month and did it in what looks like the wrong order. The prototype went online before the database existed, before payments were connected, and before anyone had agreed a price.
Specifications are a poor way to decide
Hand an owner a twelve page specification and they will approve it, because reading it properly is nobody’s idea of a good Tuesday. Then the thing gets built, they finally see it, and the real feedback starts. Everything they say at that point is correct and most of it is expensive.
What we built instead
One HTML file. All the demo data sits in the browser, so whoever opens it gets their own copy and can wipe it from the front screen. It runs on a phone, on the desk PC, and on a train with no signal.
The owner scanned an entry code, ordered a coffee, watched the order number turn green when the bar marked it ready, booked an induction, then logged a deadlift and watched a coach verify it from the desk screen. That took about ten minutes. Every decision after it was made by somebody who had actually used the thing.
Being straight about what is pretend
A demo has to be honest or it is just a trick. Ours ships with a short list of what is fake: no bank, no card machine, no door reader, and the scan buttons stand in for a reader. Prices, the menu, point values, the staff names and the induction times are sample values in one block at the top of the file, and changing them takes minutes. In the demo an order moves itself along every few seconds so one person can try the whole thing on one phone. In a real bar, someone taps the buttons.
The catch
A demo that feels finished makes people think the hard part is done. It is not. The database and the API behind every screen, the payment provider and its webhooks, the door reader, photos taken at the desk, and server-side signing so an entry code cannot be forged: none of that is in the prototype, and all of it is the expensive half. We say so on the first call, and then we say it again when the demo goes down well, which is precisely when nobody wants to hear it.
It still earns its place. We now build the demo first for anything with more than a handful of screens, whether that is a membership system, an internal tool or an assessment engine. It costs a few days at the start and it removes the argument you would otherwise have at the end. You can press the buttons on this one if you want to see what we mean.