A call-off is inventory. A forecast is not.

A forecast line on a shared spreadsheet is not coverage. It is a planning signal. Until someone ships a named quantity of a locked SKU into a named location, the bin is still empty.

For U.S.-bound medical-device components — connectors, caps, clamps, and the other plastic and accessory parts a device line runs on — two replenishment patterns close that gap: vendor-managed inventory (VMI) against agreed min/max, and call-off releases against a blanket purchase order. Both convert commercial intent into physical stock. A forecast alone does not.

We stock and distribute medical-device components and run these inventory programs for customers who need parts on the shelf, not a plan on a screen. We don't own your device's specification or its regulatory submission — that stays with your quality team. What follows is narrower: how stocking programs actually work on the floor, and how to keep call-off discipline from collapsing into forecast-as-coverage.

Start with the locked SKU — then talk replenishment

Before VMI or a call-off means anything, name the thing you're stocking: part number, grade path, and any certificate or inspection package your program already approved. Adjacent "medical-ish" inventory that "should process the same" is not a substitute. If the tote label and drawing revision don't match what quality releases, that's a quarantine candidate, not coverage.

Once the SKU is locked, choose the replenishment loop. Many programs use a blanket for commercial terms and VMI-style min/max for day-to-day top-ups. Both sides need to know which document authorizes a shipment.

VMI: the supplier watches the bin against min and max

Vendor-managed inventory is a replenishment model, not a slogan. Buyer and supplier agree the item list and min/max, share stock and consumption data, and let the supplier replenish when on-hand heads toward the minimum — typically enough to bring the location back toward the maximum. Fabrico's manufacturing VMI overview puts the loop in that order: agree scope and min/max, share stock and usage, replenish toward the max, reconcile on a cadence.

VMI names who manages replenishment. Consignment names who owns the stock until use. Programs may combine both; they are not synonyms. Mixing them up is how billing fights and "who holds the buffer" arguments start.

Three mechanics matter more than the acronym:

• Shared visibility. On-hand, open orders, and recent consumption need a single source of truth — ERP extract, portal, or agreed file — refreshed often enough that a usage spike doesn't empty the bin between updates.
• Min embeds the physical buffer, including import lead time. The minimum is the floor below which you risk a stockout. For imported parts, lead time isn't factory cycle time alone — it includes transit, brokerage, and clearance after origin handoff. Under FOB Origin, the buyer's freight clock starts at origin, so the physical minimum has to absorb that origin-to-door stretch. Raise the min when ocean or brokerage time stretches; don't leave a four-week floor on an eight-week lane.
• Max caps overstock. Without a max, "managed" inventory becomes dock theater the first time last week ran hot.

Set min and max from lead time and usage, then revisit when either changes. If the SKU or grade path changes, reset the loop — don't inherit last configuration's buffers.

ASCM's public framing is useful here: safety stock protects against forecast error and demand or lead-time fluctuation, service-level targets drive depth, and a 100 percent service level is statistically unreachable. A "safety" column that never becomes on-hand parts is theater. Size the floor from full replenishment lead time (factory plus transit plus clearance), expected usage in that window, and a fundable spike or stretch cushion. When lead times move, move the minimum.

Call-off against a blanket: the release is what moves freight

A blanket purchase order sets commercial terms — price, validity window, estimated volume, ship-to defaults, quality expectations. It does not, by itself, authorize a shipment.

Precoro's blanket-PO guide states the split cleanly: nothing moves until a release, or call-off, names quantity, date, and destination against the parent agreement. The blanket is the contract frame. The call-off is the inventory event.

That distinction is where programs fail. Teams paste an annual forecast into the blanket's "quantity" field and treat it as a standing ship order. Later someone asks why the supplier didn't just send what was forecasted. Because a forecast is not a release. Keep three numbers labeled:

• Forecasted volume. A planning signal. Not a ship authorization.
• Blanket, or commercial frame. Terms under which releases may be placed. Still not authorization until a release exists.
• Call-off, or release quantity. The firm pull — how many, which SKU, which date, which dock. This is what creates the open shipment and, eventually, the tote.

If your ERP can issue releases against a blanket, use that path. If you work by email, write the call-off as a short, structured release — parent PO, line, quantity, need-by date, ship-to — and require acknowledgment. An ambiguous "please ship some of the blanket" message is how both sides end up inventing different quantities.

Paperwork rides the release. A packing list and certificate of conformity should travel with every shipment so receiving can match SKU, quantity, and lots to the call-off at the dock — not chase identity in a later email.

How VMI and call-off fit on one program

A common working pattern: lock the SKU with quality, open a blanket for commercial terms and estimated annual volume, run day-to-day replenishment as VMI min/max, and use discrete call-offs for exceptions — first fill, engineering builds, a surge above max, or an uncovered location. Expect a packing list and CoC on every release. The forecast still matters upstream, for capacity and honest annual framing — it doesn't replace the min/max watch or the release. Updating a forecast updates a plan. A VMI shipment or a call-off changes physical inventory.

After the carton opens: keep lot identity

Inventory programs fail quietly when a supplier carton gets opened and split across bins without a new identity. The discipline that holds up on a receiving floor is blunt: treat the break as a lot event for opened units, leave the parent lot on unopened cartons, relabel opened units before they leave receiving, and carry both parent and child identifiers on the traveler. Unlabeled minutes are how identical-looking parts get mixed. Returns go to a return path with a new flag, not silently back into mixed stock. Packing list and CoC only help if the tote still carries a readable lot.

Setup checklist

• Lock the configuration. Part number, revision or grade path, and the required certificate or inspection package, before first fill.
• Choose the authorization path. VMI min/max, call-off against a blanket, or both — and name whether stock is VMI-managed, consigned, or both.
• Define min and max in units of the locked SKU. Document the calculation — demand during full lead time, including transit and clearance for imports, plus buffer — and date it.
• Separate forecast from release. Forecast lives in planning; releases and VMI shipments live in inventory transactions.
• Ship with packing list and CoC. Match paperwork to the call-off — SKU, quantity, lots — at the dock.
• Share on-hand and usage on a fixed cadence. Weekly is common.
• Preserve lot identity after a carton split. Relabel opened units before they leave receiving.
• Reconcile and retune. Count on a schedule, and when lead time, demand, SKU, or ship-to changes, revisit min/max the same week.

What we do — and don't — claim

We stock catalog components in known configurations, so standard interfaces can move without inventing lead time out of a forecast deck. VMI-style min/max, call-off against a blanket, and shared buffers keep that stock real. Packing list and CoC travel with releases. Under FOB Origin, freight timing after origin sizes into the physical minimum.

We don't invent regulatory claims for your device, own your specification, or pretend a forecast is a call-off.

Plan upstream. Stock downstream.

Forecasts still belong upstream — for capacity, commercial framing, and honest annual volume. They don't put parts in a tote. VMI against agreed min/max, sized for real import lead time, and call-off releases against a blanket, with paperwork that keeps lot identity, do.

If your program still treats a planning number as coverage, fix the authorization path first: locked SKU, named min/max or named release, shared visibility, a physical buffer, dock paperwork. Then put the forecast back where it belongs — in planning.