The Gate We Applied and Never Published

Two of our standards decide whether a product can appear here at all: a final score of 8.0, and a Trust floor of 4.0. Both have been published for months. There is a third, and until today it was in our machinery and on no page.

Before we score a product we score our own evidence about it. That is the Confidence Score, out of 100, built from five inputs: how many independent sources we found, how good they are, whether the claims trace back to anything, whether the platform ratings hold up, and whether the story stays consistent over time. Below 35 we do not publish a verdict at all, whatever that verdict would have been. Not a rejection of the product. A refusal to pretend we know.

That floor has been doing real work. Rejections in our public record were decided by it, and a reader could read the page explaining what makes products fail without ever learning it existed. It is now on The Bar beside the other two, with the count of how many rejections cite it.

Writing it down forced a second thing into the open. We had two confidence numbers and had been treating them as one rule. 35 is the floor below which nothing gets published. 60 is a different line entirely: the mark a product must clear to carry the Labs Verified badge. A product can sit between them, be published, and simply not wear the badge. Sixty one of the products live on this site are in exactly that position right now.

Conflating the two is not theoretical. One of our published rejections states a confidence threshold of 60 as though it were the rejection line, and two others name figures that do not match the rule either. The verdicts on all three still stand, because each product failed on grounds that hold. It is the stated reasons that are wrong, and correcting those records was queued as its own work with its own entry, rather than being quietly folded into this one.

The standard has not moved. What changed is that you can now read it, which is the only version of a standard that is worth anything.