Reserve pharma stock across warehouses without double promising

Two depots each promise eighty packs from an incoming quantity of one hundred. Both planners see the same supply in their local report, and neither sees the other's commitment until the truck arrives. The problem is not demand forecasting. The organisation has allowed two independent reservation decisions against one physical quantity.

Multi-warehouse inventory needs a shared commitment model for the stock being promised. Depots can retain their operating responsibilities while allocation uses a reliable view of eligible supply, active reservations and the actual transfer state. The model should explain both where stock is and who currently has a claim on it.

WHO distribution guidance provides a reference for stock records and reconciliation. WHO TRS 1025 Annex 7, section 13. The reservation design below is an engineering proposal around the organisation's approved distribution controls.

Separate four questions in the stock view

Ownership identifies the entity whose stock is being managed. Location identifies where it physically sits or which transfer currently carries it. Quality eligibility identifies the activities permitted for the relevant scope. Reservation identifies the demand against which a quantity has been committed.

These dimensions can change at different times. A transfer dispatch changes location without necessarily changing ownership. A quality hold can remove eligibility without moving the goods. A cancelled order can release a reservation while the stock remains in transit.

Represent those changes explicitly. If one field called “available at depot” combines all four, a report can accidentally count the same transfer at origin and destination or offer held stock to a new order.

Use clear labels for physical on hand, eligible and unreserved, committed, in transit and expected supply. Expected supply should carry its assumptions and confirmation state. It should not appear as an unconditional physical balance.

Resolve two competing orders against one ledger

Consider an illustrative stock position of 100 released and otherwise eligible packs. Order A requests 80 packs and Order B requests another 80. Total demand is 160 packs, leaving a 60-pack gap.

If the approved allocation policy serves A first, reserve 80 for A and leave 20 available for B. B then has a confirmed reservation of 20 and an unresolved requirement of 60. Another authorised allocation policy may divide the stock differently, but the total confirmed reservations cannot exceed the eligible quantity.

The reservation operation should check and commit against the authoritative balance as one controlled transaction. Two screens reading “100 available” at the same moment must not independently create reservations totalling 160.

Preserve the policy and priority behind the allocation. The application should not decide customer importance through an undocumented sort order. Where reallocation requires approval, record the change and its effect on the earlier commitment.

Model the transfer lifecycle

A proposed transfer is not yet stock received at the destination. Use states that distinguish request, approved allocation, picking, dispatch, transit, receipt and reconciliation as appropriate to the operation.

Link the origin issue and destination receipt with one transfer identity. Record quantities dispatched and actually received, allowing a discrepancy to remain open. A short receipt should not automatically adjust the origin history to make the transfer appear complete.

If destination planning may reserve expected incoming stock, show that commitment as dependent on the transfer and its expected eligible receipt. Recheck it when dispatch, arrival or quality information changes. Avoid counting the same quantity once as origin stock and again as destination supply within a network total.

For multi-leg transfers, keep each custody or location event connected to the same relevant population. A temporary storage stage should not create a new unrelated supply source merely because another system records it as a receipt.

Decide what a reservation guarantees

Some businesses use soft reservations for planning and firm reservations for execution. Others use a single reservation type with clear conditions. Choose the model that fits the actual commitments and explain the difference to users.

A reservation might guarantee an eligible quantity without fixing a lot until picking. Alternatively, an order may require a particular lot or shelf-life profile from the start. Those approaches have different reallocation and exception behaviour.

The FEFO allocation guide explains how lot eligibility and rotation interact. A generic reservation cannot bypass the order's relevant remaining-life, product or quality constraints. If lot selection is deferred, the system must still demonstrate that the promised quantity can be fulfilled under the agreed rules.

Show the reservation reference and its conditions wherever sales or service teams make a commitment. A dashboard estimate and a confirmed reservation should not look identical.

Replan without losing the earlier promise

A transfer delay, new hold or changed order can invalidate an earlier assumption. Identify the affected reservations and route them to the responsible planning or order-management owner.

Keep the earlier commitment and the reason for change. Automatically moving a customer date to match the latest supply can conceal a missed promise. Likewise, silently taking stock from one order to satisfy another can hide a business decision that should be explicit.

The production scheduling guide addresses upstream supply constraints. Multi-depot planning should consume realistic supply events and uncertainty rather than treating every planned production completion as immediately distributable stock.

If no feasible supply supports the original commitment, expose the shortage and required decision. A system that reports the gap early is more useful than one that fills a screen with optimistic dates.

Make cancellation and retry behaviour safe

Cancelling an order should release only its remaining active reservation. Quantities already dispatched or consumed require their own authorised return or correction processes; cancellation cannot simply add them back into stock.

A repeated cancellation message must not release the quantity twice. Retain the reservation identity and operation history so retries return the existing result. A changed cancellation request should be assessed against the current state rather than replayed as if nothing else happened.

Test the sequence where a cancellation and dispatch arrive close together from different systems. Define which transitions are permitted and how a conflict reaches an owner. Timestamps alone may not establish a reliable business order when clocks and interfaces differ.

Pilot one network boundary

Start with two depots and one representative product group. Reconcile the opening balances, transfer records and active commitments manually. Confirm that each stock population appears once in the network accounting view.

Run simultaneous orders, a partial transfer receipt, a delayed arrival, a post-reservation hold and a duplicate cancellation. For every scenario, compare eligible supply, confirmed commitments, unresolved demand and physical location.

Measure double commitments, reservation conflicts, unlinked transfer balances and the time needed to replan a changed supply event. Keep customer-priority changes visible rather than classifying them as system failures or hiding them inside an availability percentage.

The distribution software guide helps assess the broader platform. NextEdge Labs can help define the reservation contract and exception view so existing depot systems share a coherent commitment model.

Summary and next actions

Multiple depots can operate independently without making independent promises against the same supply. Preserve stock identity and transfer history while committing reservations through an authoritative allocation process.

• Reconcile physical stock, in-transit quantities and active commitments across two depots.
• Define what each reservation type guarantees and which policy governs competing orders.
• Link origin and destination transfer events without double counting expected supply.
• Test concurrent requests, partial receipts and repeated cancellations against the same balance.