A plausible delivery plan may assign more work than a vehicle can carry. A formal model makes such constraints explicit so a solver can search for or check an allocation.

The model describes variables, constraints, resource state, and possibly an objective. Its relationship to the Business Truth needs review: a mathematically valid answer can still omit something the organization cares about.

Read the result precisely

A feasible result satisfies the encoded constraints. A proven optimum is optimal relative to the stated model and objective. A proof of infeasibility concerns that model, not every possible way of approaching the real problem.

An unresolved search says less. If the budget expires before a solution is found, that does not establish impossibility. If a candidate was found but optimality was not proved, the candidate and that limitation should both remain visible.

Invalid inputs and execution errors need their own handling. They are not evidence that the business problem cannot be solved. Exact status names depend on the solver; do not collapse them into a misleading yes or no.

Decide how much evidence is enough

The Business Truth's evidence requirements determine whether a feasible plan is sufficient. If the organization requires a proven optimum, that is a stronger requirement to satisfy, not a label to infer from a promising result.

For the delivery example, two different allocations may both meet the service and capacity constraints. Either could be acceptable. Choosing between them remains a separate decision.

A solver does not decide who may make the promise.

Retain the model behind the answer

Keep the resource snapshot, model version, constraints, objective, execution conditions, and actual result status with the evidence. Without them, “the solver approved it” conceals the question that was asked.

Changes to capacity or assumptions may invalidate an earlier assessment. Append the new assessment and retain the earlier one, including the decisions it influenced.

Try again without rewriting the attempt

A larger budget, another formulation, or a different formation may be worth trying after an unresolved result. Identify the changed conditions and preserve the first attempt.

The mathematics contributes a specific form of evidence. Human review, observed outcomes, and other checks may still be necessary. The strongest useful claim is the one the recorded evidence actually supports.