Pre-registration

The practice comes from clinical research. Trials would measure fifteen outcomes, report the two that reached significance, and describe those as the goal all along. The response was procedural rather than statistical: say what you are measuring and what counts as success, publicly, before you collect data.

The business version needs four lines. What you are changing. Which number you expect to move, and in which direction. By how much and by when. And what would make you wrong.

The threshold is the line people skip, and without it any movement reads as confirmation. "Support cost per ticket down at least 8% by March 31" can be checked by somebody who was not in the room. "Improve efficiency" cannot.

The reason this works is timing rather than rigour. After the results land, a competent team can construct a plausible account of any outcome, and having constructed it, will believe it. Moving the claim earlier is the only intervention that happens before the contamination.

Once you have thirty settled predictions you can answer a question most organisations cannot: how often the team turns out to be right, and about what.