PVA — Policy Value Add

PVA — Policy Value Add — is the dollar lift the system's policy-based decisions produce above what a rule-based heuristic baseline would produce, over the same horizon, with the same inputs, against the same realized futures. Where EVA tells you what one plan is worth, PVA tells you what the policy is worth — the system that produces every plan.

Every solve produces three plan anchors. The Heuristic Baseline — what rules-based planning would have produced (the counterfactual without VYAN's optimization, against the same inputs). The Constrained MILP — what the optimizer produced under the active Decision Policy. The Ideal Unconstrained — the upper bound if every constraint were relaxed (the ceiling you're measured against; the gap to it shows where constraints are most expensive). PVA = Constrained MILP - Heuristic Baseline. The MILP is what gets committed; the Heuristic Baseline is the counterfactual; the Ideal Unconstrained is the headroom.

PVA decomposes two ways. By KPI family: how much of the lift is in customer service vs margin vs working capital vs carbon vs plan stability vs customer priority vs risk-adjusted return. By decision class: how much from pricing vs replenishment vs allocation vs scheduling vs supplier selection. The decomposition is what makes PVA actionable — the CFO sees not just that the policy is worth $3.5M, but where that $3.5M is coming from. The MIC Q3 example: $3.5M total, with +$1.4M customer service, +$0.9M margin, +$1.1M working capital, +$0.1M plan stability.

PVA accumulates continuously. Every commit's PVA contribution is tracked; the quarterly view aggregates them with full attribution. The board narrative becomes "policy delivered $X over the quarter, decomposed as Y, with confidence interval Z" — defensible, auditable, attributable. The CFO walks the audit committee through the math; the audit committee walks out with their understanding intact.