What VMI actually changes
In ordinary buying, the buyer watches its own stock, decides it is getting low, and issues a purchase order. The supplier reacts. Vendor managed inventory reverses that. The buyer opens up its stock and sales data, the supplier reads it, and the supplier decides what to send and when. Stock shows up because the supplier concluded it was needed, not because anyone at the buyer asked.
That is the whole of it. VMI is a change to one decision: who sets the replenishment quantity and timing. It is not a change to who owns the goods, who pays for them, or when. Almost every confused explanation of VMI comes from mixing that one decision up with ownership, so it is worth being blunt about the separation before anything else.
You will also see the same arrangement called supplier managed inventory (SMI), and, in searches, “vendor management inventory”, which is a misphrasing of the same thing. Continuous replenishment is a close relative. They all describe the supplier owning the replenishment decision.
Who owns the stock, and when title transfers
In a plain VMI program, nothing about ownership changes. The supplier ships, the buyer receives, title passes on delivery under whatever terms the two already trade on, and the invoice falls due on the supplier's normal terms. The buyer's cash is tied up in that stock from the day it arrives, exactly as it would be if the buyer had ordered it.
Consignment is the other thing. Under a consignment arrangement the supplier keeps control of the goods even though they are physically at the buyer's site, and the buyer is billed only when a defined event happens, usually a sale or a consumption. US accounting guidance under ASC 606 treats an arrangement as consignment where the supplier still controls the product until such an event, can require the product back or send it elsewhere, and the holder has no unconditional obligation to pay for it. In that case the supplier does not recognize revenue on delivery. PwC's summary of consignment arrangements sets out those indicators; your accountant should be the one applying them to a specific contract.
The two are independent. You can run VMI on stock you own outright. You can hold consignment stock that you still reorder yourself. Retail programs often run both at once, which is where the confusion starts. Scan-based trading is the retail version of consignment: the supplier owns the goods on the shelf until the register scans them, and title passes at that moment.
VMI vs buyer-managed replenishment vs consignment stock
The table separates the decision from the ownership. Read the first two rows together and the rest follows from them.
| Buyer-managed replenishment | Vendor managed inventory | Consignment stock | |
|---|---|---|---|
| Who decides the reorder | The buyer, from its own forecast and reorder points | The supplier, from data the buyer shares | Either. Consignment says nothing about who decides |
| Who owns the stock on the shelf | The buyer, from receipt | The buyer, from receipt, unless the contract also makes it consignment | The supplier, until the triggering event |
| When title transfers | On delivery, under the agreed shipping terms | On delivery, under the agreed shipping terms | On the defined event, usually sale or consumption |
| Who carries the cash | The buyer, from the day stock lands | The buyer, from the day stock lands, on whatever payment terms already apply | The supplier, until the stock is sold or used |
| What data is shared | Purchase orders and forecasts only | On-hand, units sold, on order and in transit, by item and location | At minimum, usage or sales, so the supplier can invoice correctly |
| Who it suits | Buyers with good demand planning and few suppliers that matter | Repeatable, steady-demand items where the supplier plans better than the buyer | Expensive, slow-moving or unproven items the buyer will not fund up front |
What data gets shared
A supplier cannot plan your stock from your stock level alone. Two stores with 40 units on hand are in completely different situations if one sells three a day and the other sells thirty. So the minimum useful feed is:
- On-hand quantity by item and by location, with a stated as-of time.
- Units sold or consumed in the period, which is what gives the supplier a rate of draw.
- On order and in transit, so the supplier does not count the same replenishment twice.
- Returns and adjustments, which otherwise make sales look better than they were.
- The agreed targets: minimum and maximum levels, or a days-of-supply target, per item and location.
Most established programs move this over EDI. The transaction set specifically built for it is the X12 852 Product Activity Data, whose stated purpose is to advise a trading partner of inventory, sales and other product activity information. Related sets in the same X12 catalog do the rest of the job: 846 Inventory Inquiry/Advice for stock positions, 850 Purchase Order and 855 Purchase Order Acknowledgment for the order itself, 856 Ship Notice/Manifest for the shipment, and 810 Invoice for billing. Smaller programs skip EDI entirely and run on an API feed, a shared dashboard, or a scheduled file. The format matters much less than the cadence and the accuracy.
Accuracy is the part that quietly kills VMI programs. If your counts drift, the supplier is planning against a number that is not true, and it will be your shelf that is wrong. Regular cycle counting is not optional once someone else is acting on your numbers.
How replenishment is triggered
The common mechanisms are ordinary inventory control, applied by the supplier instead of the buyer:
- Minimum and maximum levels. The supplier tops the item back up to the maximum whenever it drops to or below the minimum. The simplest arrangement and the easiest to audit.
- A reorder point. A threshold set from average daily sales, the supplier's lead time, and a buffer of safety stock. Crossing it triggers a shipment.
- A days-of-supply target. Rather than a unit count, the agreement fixes a cover period, say two weeks of expected sales, and the supplier ships whatever keeps the location at that cover as demand moves.
- A forecast the two parties agree. Larger programs run this as collaborative planning, where buyer and supplier reconcile one demand forecast and the supplier replenishes against it.
Whichever is used, the agreement should also pin down the things that go wrong: a maximum the supplier may not exceed, what happens to stock that does not sell, who pays for expiry or obsolescence, how fast the targets can be changed, and what the buyer can do if service slips.
What VMI is actually good at
Fewer stockouts on predictable items. The supplier sees the draw earlier than a purchasing cycle does, and it is the party best placed to know its own lead times and constraints.
Less buyer admin. Routine reordering is a large amount of low-value work. Handing it over frees a small team to plan the items that genuinely need judgment.
Better supplier forecasting. This is the real prize, and it belongs to the supplier. Seeing actual consumption instead of only the orders placed against it damps the amplification of demand swings up the chain, the bullwhip effect that Lee, Padmanabhan and Whang analysed in 1997. A supplier that can see through to real demand can smooth its own production and hold less buffer.
What it costs you
You lose control of a decision that spends your money. Unless the stock is on consignment, every unit the supplier decides to send is your cash and your storage. A supplier with a quota has an incentive to keep you full. Caps and a maximum are what keep that honest.
Dependency. Once the supplier holds the replenishment logic, the target settings and the history, moving to a different supplier means rebuilding all of it. That is real switching cost, and it shows up in your next negotiation.
The supplier has to be able to do it. VMI moves planning work, and often inventory risk, up the chain. A supplier without the systems, the staff, or the appetite for that will either do it badly or price it in. Plenty of small suppliers should simply say no, and a buyer pushing VMI onto one that is not equipped usually gets a worse result than it started with.
It fits fewer items than people expect. VMI works on repeatable, reasonably steady items with a real history. It does not fit a new launch with no demand signal, a seasonal spike the supplier cannot see coming, or a catalog that churns.
Where a 3PL fits
Mostly as the source of truth, not as the decision maker. If your stock sits in a third-party warehouse, that warehouse is where the on-hand number comes from, and its warehouse management system is what makes the number credible. A 3PL in a VMI program typically receives the supplier's shipments, records them by item and location, reports on-hand and outbound movement on the agreed cadence, runs the cycle counts that keep the record honest, and ships to the plan it is given.
What a 3PL does not normally do in a VMI program is make the reorder decision. That decision is commercial, and it sits between you and your supplier.
We run one warehouse, in Austin, Texas, and in a VMI arrangement this is the role we play: accurate counts and regular reporting for brands whose suppliers or wholesale buyers need to see the numbers. If you want help setting reorder points and buy quantities yourself, that is a separate service, inventory planning, and you approve every purchase it recommends.
Is it worth doing?
Ask three questions in order. Does the item have a steady, observable demand history, or are you asking someone to plan noise? Can you hand over counts that are actually right, refreshed often enough to act on? And is your supplier equipped to plan, or are you about to move a job to someone who will do it worse than you do?
Three yeses and VMI usually pays for itself on those items. One no, and you are better off keeping the decision, tightening your own inventory management, and revisiting it when the answer changes. Terms used on this page, including reorder point, safety stock and EDI, are defined in our logistics glossary.