The cost of building the wrong thing

Ask a team how much it cost to build the feature that didn't work, and you'll usually get an answer in engineering weeks. Six weeks of one engineer. Three weeks of two engineers and a designer. The kind of number that fits on a slide.

That number is almost always wrong, and it's almost always wrong in the same direction. It's too small.

The direct cost — the time it took to build the thing — is the only layer that gets measured. There are two more, and the one nobody talks about tends to be the largest.

The opportunity cost.

Whatever you built, you didn't build something else. If the bet had worked, that opportunity cost would be paid by the gains. If it didn't, you're left holding the bill alone. For most teams, the better thing they weren't building is invisible — you can't easily count what didn't happen — and so this cost rarely shows up in postmortems. It's there anyway. It just doesn't have a line item.

The compounding cost.

This is the one. Once a bad bet ships, it doesn't sit politely in the corner. It becomes part of the context that shapes every decision after it.

The next quarter's roadmap is now written assuming the wrong thing exists. The next architectural decision is now made assuming the wrong thing exists. The next hire is onboarded into a codebase that contains the wrong thing — and inherits a slightly distorted picture of what the product is for. Three quarters later, you're making decisions that are warped, faintly, by a bet you'd long since stopped thinking about.

The compounding cost is invisible per decision and enormous in aggregate. It's the reason that mature products feel hard to change and young ones feel easy. It's also the reason that “we'll just rip it out later” almost never happens — by the time you'd want to, the rest of the system has grown around it.

We think this is the right frame for the “move fast” conversation.

The usual framing pits speed against quality. That's not quite the right axis. The right axis is speed against reversibility. A fast bet that you can undo cheaply if it's wrong is cheap regardless of outcome. A fast bet that compounds into the system whether or not it works is expensive even when it works, because you've now committed to it indefinitely.

The implication is uncomfortable for anyone who's spent the last decade absorbing the “ship fast” gospel: speed without reversibility is not leverage. It's debt — the slow, quiet, hard-to-attribute kind that's still on the balance sheet years later.

Two operating modes survive this analysis. You can build the right thing slowly. Or you can build the wrong thing reversibly. The combinations that don't work — slow and irreversible, fast and irreversible-and-wrong — are the ones that quietly sink companies, often without anyone naming the cause.

The frame we keep returning to: every product decision is borrowing from a future budget. The interest rate is set by how reversible the decision is. The principal is set by how much downstream context the decision creates.

You don't get to skip the bill. You only get to choose, in advance, how big you want it to be.