Order Fulfillment Priority Decision

There are no documented fulfillment priority rules. When inventory is short, whoever shouts loudest gets their order shipped first. A VP calls the warehouse and says 'ship Acme before anyone else' and that becomes the priority — until another VP calls with a different instruction. There's no written logic for who gets product first.

AI cannot allocate orders because no priority criteria exist. Every allocation decision requires human intervention and political negotiation.

An informal priority system exists — 'strategic accounts first, then by due date' — but it's not written down and varies by who's making the call. The warehouse manager prioritizes based on what they were told last, not a consistent policy. During shortages, the sales VP, operations VP, and CFO each have different priorities, and the allocation meeting becomes a negotiation rather than a decision framework.

AI could sort orders by due date, but cannot make meaningful priority decisions because the criteria are informal, contradictory, and politically driven rather than systematically documented.

A written fulfillment priority policy exists — customer tier rankings, SLA-based due date priority, and escalation paths for conflicts. The policy is documented in a procedures manual. But applying it requires manual judgment — someone reads the policy and applies it to the current order backlog. The policy doesn't cover every edge case, so exceptions still go to meetings.

AI can sort orders according to the documented priority criteria, but cannot handle edge cases or multi-factor trade-offs because the policy is documented as prose guidance rather than computable rules.

Fulfillment priority rules are encoded in the order management system. Customer tier scores, SLA deadlines, order margin weights, and relationship risk factors combine into a computable priority score. When inventory is constrained, the system calculates the optimal allocation and presents it to the order manager for confirmation. The 'allocation meeting' becomes a review of the system's recommendation rather than a from-scratch negotiation.

AI can calculate optimal order allocation based on encoded priority rules and present recommendations. Cannot yet autonomously allocate because exception handling still requires human judgment for high-stakes trade-offs.

The fulfillment priority model is a formal ontology linking priority rules to customer contract SLAs, order margins, relationship risk scores, production capacity, and inventory positions. Priority calculations account for cascading impacts — allocating to Customer A means delaying Customer B, which triggers an SLA penalty. An AI agent can evaluate 'what is the total financial impact of every possible allocation scenario for this week's constrained orders' and present the optimal outcome.

AI can autonomously allocate orders in most scenarios — optimizing across customer value, SLA compliance, margin impact, and capacity constraints. Human intervention needed only for genuinely novel situations not covered by the model.

Fulfillment priorities calculate themselves in real-time from continuous operational signals. Inventory changes, production completions, customer urgency signals, and capacity fluctuations all feed into a continuously optimizing allocation engine. There is no 'allocation decision' — there is a continuously updating optimal allocation that responds to every operational event as it happens. The allocation is always the best possible answer given current reality.

Fully autonomous order allocation. AI continuously optimizes fulfillment priorities in real-time based on the complete operational context. No allocation meetings needed.

Ceiling of the CMC framework for this dimension.