Dozens of SKUs from the Same Supplier — At What Frequency Do I Order Each So Trucks and Stock Cost Are Minimum Together?

Editor’s note

In plain speech this problem is called “consolidating supplier orders”, “filling the truck”, or “shared order day”. In academic literature it sits under one umbrella: the Joint Replenishment Problem (JRP) — a classical OR family developed since the 1970s.

This problem is about multiple SKUs sharing a joint order cycle instead of each SKU running its own (s,S). Major setup costs (truck, customs, contract) are split across the N products. The (s,S) model (problem #004) solves reorder-point + order-up-to-level for a single SKU; Wagner-Whitin (problem #016) solves single-SKU multi-period lot sizing; newsvendor (problem #011) solves single-SKU single-period under demand uncertainty. JRP is a different family: N SKUs sharing a major setup.

Most-skipped point in the sector: power-of-two policies. In practice each SKU renews at its own period; if the period values are picked arbitrarily, the order calendar becomes chaotic — it’s unclear which SKUs arrive in any given week, customs-filing pooling falls apart. The academic result is that restricting periods to powers of two of the base period T (T, 2T, 4T, 8T, …) keeps the solution within 2% of the true optimum and adds schedulability: every SKU is guaranteed to be on the order list at integer-multiple times of T, truck-fill becomes predictable. The principle extends beyond JRP into one-warehouse-multi-retailer and broader multi-echelon coordination.

Second skipped point: deterministic vs stochastic. The classical JRP model assumes constant demand; real retail demand is random. The stochastic extension is the can-order policy: when one SKU hits its reorder point the joint order is triggered; SKUs currently below the can-order threshold (s < c < S) are added as joiners. The approach is common in practice but analytically complex; for small-to-mid retail a simple power-of-two + safety-stock blend is usually enough.

Step-by-step — for the SMB

Stage 1 — Measure first, plan second. At least 12 months of per-supplier data: which SKUs ship together in each joint order and at what quantity, truck-fill percent, SKUs per customs filing, weekly sales and variance per SKU, planned vs actual lead time per SKU, supplier MOQ value and violation count. Demand correlations: which SKUs rise together, which substitute.

Stage 2 — Extract the knowledge capital. Supplier-SKU matrix (per supplier: active SKU count, shared major setup cost, average order frequency, MOQ rule). Within-supplier ABC: which SKUs are high-value (must come every T), medium (every 2T), low (every 4T). Separate major vs minor setup: truck haul + customs filing fixed (major), SKU-specific label + QC variable (minor).

Stage 3 — Pilot. 8-12 weeks. For a supplier set (e.g. the 1-2 suppliers with the highest SKU count) fix base period T (e.g. 2 weeks). Compute power-of-two multiplier m_i per SKU (high-value m_i=1, medium 2, low 4). Run in parallel with the current independent (s,S). Success criteria written in advance: truck count -10% minimum, total order + stock cost -5% minimum, MOQ violations -50%.

Stage 4 — Rollout. 9-15 months to extend across all main suppliers. Stochastic extension (can-order or safety stock + periodic review) can be added. Quarterly operations committee: truck-fill, MOQ compliance, stock turn, T and m_i updates at supplier-contract renewals.

Risks — what can go wrong

1. Unmodelled demand correlation breaks the plan. If two SKUs substitute (when one sells the other stalls), the T and m_i values computed under the independence assumption come out wrong in the field. Recalibrate the correlation matrix every 6 months.
2. Soft-modelled supplier MOQ. Academic JRP rarely models MOQ; if ‘add-on SKU’ suggestions are not automatic for cycles below MOQ, the operation drifts into forced add-ons or pays the high small-order surcharge. MOQ must be a hard constraint.
3. Unaccounted customs-filing pooling. Opening one customs filing is a meaningful part of major setup; the joint order cycle must produce the correct filing (HS-code compatibility, single declaration with SKU list). This is operational discipline that must be checked before contract renewal.
4. Single-supplier lock-in. If the JRP optimum concentrates on one main supplier, a price hike or delivery slip there leaves the operator weak. Balance the JRP optimum at the policy level with a multi-supplier rule (every SKU has at least two approved sources).

Technical view of the solution method

Objective function choices:

• Objective 1 — Minimum total annual cost (major setup + minor setup + holding): Classical JRP; for cash-tight operators.
• Objective 2 — Minimum truck/container count: For logistics-capacity-bound operators.
• Objective 3 — Minimum cost under service-level constraint: For customer-facing operators.
• Objective 4 — Maximum stock turn: For working-capital-bound operators.

Multi-objective: weighted sum or hierarchical (first MOQ compliance, then truck count, finally holding cost). If a power-of-two policy is picked, schedulability is an extra objective.

Academic sources

Listed in the sources block of this page’s frontmatter.