Getting it right the last time

Yesterday I spent part of the day correcting records that have been wrong for years. Not catastrophically wrong — the business ran, money moved, customers were served — but wrong in a way that would eventually matter. Payment entries tagged to the wrong account. Receipts pointing at the wrong place. Small mislabellings compounding quietly in the background while everything else looked fine on the surface.

Nobody noticed. The books balanced near enough. And that's exactly the problem.

There's a version of founder discipline that looks dramatic: the pivot, the kill decision, the hard conversation. But most of the actual work looks like this — going back into something that already shipped, already runs, already earns, and making it correct. Not for today. For the next time someone has to trust it.

The first time is rarely the last word

Andrew Hunt and David Thomas have a line I keep coming back to: it's not at all important to get it right the first time — it's vitally important to get it right the last time. That sounds like permission to be sloppy. It isn't. It's a more demanding standard than getting it right the first time, because it means you have to stay in the game long enough to fix it, notice it needs fixing, and actually go back.

Most founders I know are allergic to going back. Forward feels like progress. Backward feels like failure. So the quiet errors accumulate — in the records, in the process, in the logic — until one day they aren't quiet anymore.

The solo operator's version of this is especially uncomfortable, because there's no team to blame and no handoff to hide behind. When something has been wrong for four years, you were the one who let it sit. That's the part you have to be honest about before you can fix it without resentment.

I've noticed I do my best correction work when I stop calling it a mistake and start calling it a first draft. The first implementation was a guess made under pressure with incomplete information. Yesterday's work was the revision. Both are necessary. Neither cancels the other out.

Running a small business is mostly revision. The product you launched isn't the product you'll keep. The process you set up in year one isn't the process that will carry you to year five. The records from before you understood what you were building need to be brought into alignment with what you actually built.

Correction isn't failure. It's the work that makes the rest of it load-bearing.

The only bad version is the one you never go back to fix.