Vendor-Managed Inventory (VMI): When It Works and When It Doesn't

Under VMI (vendor-managed inventory), instead of issuing purchase orders the customer shares its sales or consumption data and stock levels with the supplier, and the supplier decides how much to ship and when. The customer still owns the warehouse space, but the supplier is responsible for keeping the agreed stock level.

The economic logic of VMI is visibility. Instead of seeing batched, irregular orders, the supplier sees real consumption. This is the core remedy for the bullwhip effect: when order variability approaches actual consumption, production and shipment planning get smoother for the supplier and both stockouts and excess inventory fall for the customer. Shipments can also be consolidated and optimized.

But VMI does not work in every setting. Suitable conditions are fairly regular, predictable consumption; a limited set of key items; a long, trusted relationship; and the customer's ability to provide accurate stock and consumption data. For highly volatile or project-driven items, a supplier without sight of the customer's future plans must hold high safety stock and reflect its cost in the price.

The first implementation step is agreeing rules, not tools. Minimum and maximum stock at the customer site, a precise definition of a stockout and its financial consequence, inventory ownership (the supplier's until consumption, or the customer's from delivery), and the method for resolving data disputes should all be in writing. Without these rules, when a stockout occurs each side blames the other.

Inventory ownership matters more than it seems. If stock is sold to the customer before consumption, VMI has only moved the decision, and the inventory capital still sits with the customer. In VMI with consignment stock, inventory capital moves to the supplier, which is more attractive to the customer, but the supplier will want a different price or payment terms to accept it.

A common mistake is treating VMI as a way to shed planning responsibility. The customer still has to give the supplier its forecast and upcoming plans (promotions, maintenance shutdowns, product changes). Another mistake is starting with all items. It is better to pilot with 5 to 10 stable items and one trusted supplier, and measure the result.

A practical starting point: choose one key supplier with a good relationship, identify its stable items, and measure three metrics before and after the pilot: stockout rate, average inventory (days of cover) and freight cost per unit. If at least two do not improve within two or three months, either the relationship or the item selection was not right.