Four things must be real
The recurring decision needs an accountable operator and cadence. Its evidence must be joinable at the required cutoff. The organisation must control a specific action. The result must have a measurable financial or operational consequence.
If one of those conditions fails, a more sophisticated model does not rescue the opportunity.
A QOA is allowed to say stop
A paid QOA maps the decision and systems, tests evidence viability, defines the baseline and counterfactual, and identifies a bounded pilot. It should expose unjoinable identities, missing authority, weak labels and non-recognisable value before implementation.
The commercial output is therefore a decision: proceed to a pilot, redefine the scope, or stop.
What remains unproven
A QOA does not establish production lift. That requires a tenant-calibrated champion, governed shadow evidence and a measured outcome design. Nor does it guarantee source access, adoption or finance recognition.
Being precise about the boundary makes the pilot more credible and protects both parties from a benefits case built on assumptions.