Each item is scored from one to five, and every item carries a written description of what a 1, a 3 and a 5 look like. A 2 clearly passes 1 but does not fully meet 3; a 4 fully meets 3 and shows some, but not all, of 5. When a scorer is torn between two numbers, the lower one is taken and the reason is written down.
Every score needs evidence. A link, a screenshot or a search result with its date sits next to every number. A score without evidence is not a score. What a friend of the subject would say does not matter; what matters is what a stranger can find.
When nothing can be found. Most items describe things the public can see. For those, no evidence means a score of 1: in an external read, absence is the finding. A smaller set of items describes what happens inside a company, such as retention, pipeline and acquisition cost. Those are scored from the nearest public sign, with the confidence marked low; where there is no sign at all they score 3 and are flagged as not observable, because a company is not punished for what the public simply cannot see.
Unanswered is not the middle. In the self-assessment, an item left unanswered scores 1, never a middle value. An untouched facet reads low rather than average, and the report says so.
Every facet carries a confidence level: high when most scores rest on direct public evidence, medium when many rest on nearby signs, low when most rest on inference. Demand is almost always low in an external read, and the read says that plainly.