The switches you leave off
Building with care isn't timidity — it's how you find out what you were about to get wrong.
Somewhere in the middle of building a loyalty program yesterday, I almost fixed a problem that didn't exist.
The data said roughly a thousand customers were missing — no orders on record, no tier assigned, unreachable by the campaign I was putting together. I was about to route around them. Treat them as phantom records, a data artifact, write them off. Then I looked more carefully at the underlying numbers.
Almost all of them had ordered. Most held the highest-value discount tags in the entire system. They were not missing customers. They were the best customers. The counter that said zero was a placeholder — hardcoded, not measured. A sentinel value masquerading as a result.
If I'd moved faster, I would have spent real effort solving a fake problem, and the campaign would have launched without its most important recipients.
There's a quote that gets passed around in startup circles. Patton said a good plan violently executed now beats a perfect plan executed next week. People use it to justify speed. I've used it that way too.
But I think the part that gets dropped is that Patton was talking about military command, where the cost of hesitation is someone else occupying the ground. In a one-person business, the ground mostly waits for you. The cost of hesitation is usually lower than the cost of a bad assumption shipped fast.
Yesterday I built several interconnected systems — a delivery layer, a minting engine, a routing change for a tool that reads incoming messages — and left every trigger switched off. Nothing fired. Nothing reached a customer. The whole day was building toward a moment that hasn't arrived yet.
That used to feel like not finishing. Now it feels like the actual work.
The switches are off because I'm not certain yet. Not because I'm afraid, but because I'm measuring. The difference matters. Fear keeps the switches off forever and calls it prudence. Discipline keeps them off until you know what you'd have gotten wrong — and then flips them.
Yesterday the thing I would have gotten wrong was a thousand customers — the best ones — excluded before anything went out.
The plan is still violent. It just isn't time yet.