Why it matters when you are writing
The workable test has two parts. Can the fact be found, uncited, in several independent reference works or textbooks in the field? And would your intended reader accept it without wanting to check? Both yes, and you can leave it unattributed. Either no, and cite. This is a judgment, and judgments are supposed to be made deliberately rather than by default.
Audience does most of the work. That the Berlin Wall fell in 1989 needs no source anywhere. That a particular protein folds through a specific intermediate is common knowledge in a structural biology lab and needs a citation in a general science essay. Writing for an interdisciplinary journal usually means citing more than you would in your home subfield, because your reader’s baseline is different.
One category is not covered by the exception no matter how familiar it feels: specific numbers. A prevalence rate, an effect size, a survey percentage, a measured constant — these come from a study, and the study gets cited even when the figure circulates widely. If your sentence contains a digit that somebody had to measure, it needs a source.