The removal test: how to tell which of your rules are actually load-bearing

Every founder has at least one rule that exists purely to make them feel safe, not to make the work better. A report nobody quotes back to you. A CC that only gets opened once something's already gone wrong. A sign-off that has never once said no.

How do you find which ones?

Run the removal test. Pick one recurring check: a report, an approval step, a status meeting. Stop requiring it for two weeks. Don't explain why, don't frame it as an experiment, just stop asking for it, and watch what actually happens to the work.

What does the removal test usually show?

Most of the time, nothing happens. The work continues at the same pace and the same quality. Nobody escalates. Nothing breaks. That is the result, and it is the useful one: the rule was never catching bad outcomes, it was just giving you visibility, and visibility felt enough like control that nobody questioned it.

Occasionally something does slip in those two weeks. Good, that means the rule was genuinely load-bearing, and now you know exactly what it protects against instead of "it's just how we do things here." Put it back, on purpose this time, with the specific failure it prevents written down next to it.

Why does "nothing happened" matter more than it sounds like it should?

Because every rule that survives only because nobody tested it is still costing you. Someone still has to produce the report. Someone still has to wait for the sign-off before moving. And the person doing the work gets a clear, if unspoken, message: you are trusted to follow the steps, not to own the outcome. That is an odd thing to communicate to someone you hired specifically for their judgment.

Run this across ten or fifteen recurring checks and a pattern shows up fast. In most companies we've looked at, somewhere around a third of them turn out to be pure visibility, kept out of habit rather than necessity. Removing them doesn't just save hours. It hands the outcome back to the person actually doing the work, which tends to improve it.

How to run this yourself

List every recurring check in your company that exists to keep you informed: reports, sign-offs, CCs, status meetings. Pick three. Quietly stop requiring each one for two weeks. Write down what actually happens versus what you expected to happen. Whatever survives untouched wasn't protecting anything. Whatever a stakeholder pushes back on immediately was real, so put it back, and now you know precisely why it's there.