Four conditions must hold
The decision needs a named operator and cadence. Its historical records must be joinable as they existed at the cutoff. The organisation must control a specific action. The result must have a measurable financial or operating consequence.
If any one of those conditions fails, a more complex model will not rescue the opportunity.
A valid result may be to stop
The opportunity assessment maps the decision and systems, tests the historical record, defines the baseline and specifies a bounded pilot. It should reveal unjoinable identities, absent authority, weak outcome labels and value that finance cannot recognise before implementation begins.
Its commercial output is a decision: proceed, change the scope or stop.
What remains for the pilot and production
The opportunity assessment does not prove production improvement. That requires a client-calibrated model, an approved shadow comparison and a design for measuring outcomes. It also cannot guarantee source access, operator adoption or finance recognition.
A precise boundary produces a stronger pilot and prevents the business case from resting on assumptions.