A system whose cost you don't know isn't an investment. It's a habit.

In most of the companies I’ve seen up close, the hardest line of spending to reconstruct isn’t the big one. It’s the one that accumulated in small pieces. Monthly subscriptions, services switched on for a project and never switched off, tools somebody trialled that simply stayed.

Nobody knows what they cost together. Plenty don’t know who still uses them.

The problem isn’t the figure, which is often bearable. The problem is that spending you don’t look at stops being a decision. It carries on out of inertia, and inertia is never neutral: it’s saying something about your work, except you didn’t say it.

This got much sharper once tools appeared whose cost depends on how much you use them. With a flat subscription you can afford not to look: you know what you pay. With something that consumes on every operation, not knowing what it costs means not knowing anything about what you’re doing.

A few months ago I spent a week building this kind of accounting for the systems I’d put together. Not an estimate: real spending, split by piece, by task, by time of day. It was less painful than I expected, and what I found wasn’t what I’d gone looking for.

I was looking for waste. I found a hierarchy.

The part that runs every fifteen minutes, all day, every day, was doing simple work and cost almost nothing per run. The part that wakes up occasionally, to decide something that actually matters, cost far more per call but ran rarely. Added up over a month, the second was a fraction of the first.

Out of that came the only useful decision: put the cheap tool where volume is high and judgement isn’t needed, and the expensive one where judgement counts and volume is low.

Written down it sounds obvious. It isn’t, because without the measurement people do the opposite. They pick the best tool once and use it for everything, which is the simplest way to pay a lot and not notice.

Part of this argument holds well outside technology, and it’s why I bother with it. Every recurring cost you don’t look at is an opinion about your work that you never expressed. You’re saying that thing is worth what it costs, every month, without ever having checked.

And when someone asks why you use it, the honest answer is usually that it was already there.

The question I use to get out is blunt and takes ten minutes. If this stopped working tomorrow morning, how long before anyone noticed?

The answers split into two clean groups, and one of those groups is a list of things to cancel.