The most dangerous output is one that reaches a conclusion the evidence doesn’t support.
In regulatory work, this was the pattern behind most serious failures. The data was there. The problem was the jump. An analyst with incomplete information who still needed to deliver something. A system that always had to answer, because returning nothing looked like a failure.
I am building the opposite principle into ZoneReport, a side project I’m working on. Every report requires a minimum of five data layers out of eight. Below that threshold, no report ships. The product refunds and stops. A confident wrong answer does more damage than an honest gap.
Scoping your conclusions is not weakness. It is what makes the output worth something.
Saying “insufficient evidence to conclude” is what separates analysis from noise.