Value tells us whether it matters. Readiness tells us whether we can act. Trust tells us whether we should act in this way. Evidence tells us how confident we should be. A single number would hide all four.
ValueStrongIs this worth solving? Consequence, directional exposure, who it touches, strategic weight. Frequency is an input, never a gate.
ReadinessReady with conditionsCan we act on it? Process and Data are the foundation. People and Capacity are prerequisites written into the charter.
TrustRequires reviewShould we act this way? Read first as a posture from the problem itself, then in full against the proposed intervention.
EvidenceModerateWhat do we actually know? Each input is counted, observed, reported or unknown. Here frequency and process are counted; the cost per occurrence is unknown.
Example signal, for illustration. Manageable never means a technology is safe; it means the identified trust requirements can be controlled within the proposed intervention. The same problem can read Manageable for a process redesign and Requires review for an AI voice system, which is why Trust is read twice.
The reversibility spectrum
The question that separates two technically similar uses is not "can technology do this" but "what happens if it gets it wrong, and how soon would anyone know." This is why Assist is often preferable to Automate: not because people are better at the task, but because the cost of an undetected error differs.
Easy to reverseA draft a person edits
RecommendationA suggestion a person accepts or ignores
TriageRouting or sorting a person then performs
Human-approved actionPrepared by technology, released by a person
Automated, detectableA reminder, a classification, a routed invoice
Consequential decisionChanges what a person receives or is owed
Hard to reverseA filing, a payment, a termination