Evidence tier
A label for where a claim's evidence came from, so you can weigh it before you act on it.
An evidence tier says where a number came from, which is usually more important than the number itself. Acumin uses four, in ascending order of authority: public data anyone can see, first-party analytics from your connected accounts, paid-platform results, and outcomes from a system that knows whether the thing actually sold.
The point of labelling is that two identical-looking claims can rest on completely different foundations. "This format outperforms" derived from public view counts is a weaker statement than the same sentence derived from your own retention data. But on a slide, they look the same.
Tiers are about provenance, not certainty. A well-sampled public-data finding can be more trustworthy than a first-party finding drawn from four posts. The tier tells you what kind of thing you're holding; the sample size and the confidence band tell you how hard to lean on it.
Where this comes up
Four tiers of content evidence, from public view counts to sales data: what each can prove, what it can't, and how to know which one you're holding.
The difference between a number a system counted and a number someone told you, and why almost every content report quietly mixes the two together.