A pattern found after searching everywhere deserves a different confidence from one predicted before looking.

Where you look changes what a finding is worth. A pattern spotted after combing through everything was drawn from a huge number of chances for coincidence, while the same pattern named in advance and then confirmed had only one. The evidence looks identical and it is not.

So record your expectations before you check. If you are testing a change at work, in your health, in a relationship, say beforehand what you expect and what would count as failure. And when someone shows you a pattern in data, ask how many other patterns they examined before this one caught their eye.