Situation
A diagnosis exists and someone is tempted to summarize it as a grade ("our marketing is a 6/10") or to treat Unknown areas as failures.
Outcome
You can separate observed evidence, insufficient evidence, and recommendation, and you can explain why there is no score.
Steps
- 1Open the Diagnostic and pick three capability areas: one Observed, one Unknown, one you're unsure about.
- 2For the Observed one: write down the evidence and the confidence (high / medium / low). Then note the run's evidence basis, which applies to the whole report: a direct-scan-only run supports narrower claims than a grounded read. Say so.
- 3For the Unknown one: write down what "Unknown" means here ("could not verify enough") and name the next verification (connect an integration, add a source, correct a profile).
- 4Practice the sentence: "Unknown is not zero. Missing evidence is not proof of poor performance. Scraper coverage is not a business grade."
- 5Check that every claim has a citation. Model-derived claims without source IDs don't enter customer views. If you see unbound prose, treat it as a draft, not evidence.
- 6Decide: is a reassessment warranted? Reassessment reports what changed (new, unavailable, or unchanged evidence), never score deltas.