Where Should the Next Book Wait?
Inventory placement, reserve stock, risk pooling, replenishment and transfers.
Part 13 of 18 · Before the Shelf
Enough books, wrong place
Consider a separate placement example: eComm Inc owns 1,000 copies, but the facility that could fulfill my order today has none. Most copies sit in a bulk warehouse whose next replenishment arrives tomorrow. Counting the network has answered how many books exist without answering which promise those books can support.
Inventory placement decides which future fulfillment choices will exist. It therefore belongs in the inbound story. Choosing where Maya's load first enters the network shapes later transport, availability and capital requirements.
Depth, breadth and response time
Putting the first copy of a title in a local facility broadens its assortment. Adding another copy deepens protection against demand for that title. A slow-selling book can still deserve local stock when remote service is expensive or too slow. Conversely, duplicating every title in every location can strand a great deal of capital.
Reserve stock held upstream can serve several downstream facilities. Forward stock sits closer to expected use. The reserve can pool uncertainty, but it cannot protect a promise faster than the replenishment path allows. Two hours of local demand cannot be served from a reserve that takes two days to reach the shelf.
Pooling is conditional, not magic
Here is a deliberately small probability example. Two regions each have daily demand standard deviation of eight units. Assume independent demand between regions and days, a two-day protection interval and a two-standard-deviation buffer rule. Separate buffers total approximately 45.25 units: two times 2 × 8 × √2. A perfectly shared pool under the same response-time assumption needs 32 units: 2 × √(2 × 8² × 2).
The apparent reduction is about 13.25 units before rounding. It depends on the assumptions. If regional demand moves together, the pooling benefit shrinks. If central storage adds response time, more protection may be required. This original calculation illustrates the risk-pooling concepts taught in MIT's public inventory materials; it is not a measured saving or a service guarantee.
Replenishment is not the same as rebalancing
Replenishment restores stock against expected consumption. Rebalancing moves existing inventory from an excess location to a deficient one. An order-specific transfer may serve a particular commitment. The vehicle can carry all three, but their urgency and decision rules differ.
A transfer should not create stock in two places. At departure, the origin's on-hand quantity falls and in-transit quantity rises. At arrival and valid receipt, in-transit falls and destination stock rises. Eligibility may change later. Reservations are commitments against these balances, not additional physical goods.
Multi-echelon thinking
An echelon is a level in the supply network. If the bulk warehouse treats every forward facility's safety buffer as independent new customer demand, it can protect against the same uncertainty repeatedly. Coordinated planning considers downstream inventory and demand together while preserving the local protection needed during transit.
Capacity and ownership restrictions remain real. Available space in the wrong temperature zone is not suitable space. Another seller's stock is not freely interchangeable with Maya's. A promising allocation must satisfy both physical compatibility and commercial eligibility.
A smaller buffer needs a credible response path
The pooled calculation saves about 13.25 units only under the stated demand and response-time assumptions. Before moving stock centrally, test the route back to each point of demand. If centralization adds a day, both demand during that extra day and its uncertainty need protection. Correlated demand can consume more of the apparent benefit.
eComm Inc therefore needs two views together: where stock is and how soon it can become usable somewhere else. A reserve may be ideal for slow demand that can tolerate the replenishment interval. A nearby forward location may be necessary for promises the reserve cannot reach in time.
Maya's destination instruction should follow that policy. Sending every load to the cheapest receiving site can leave a network full of books that need another expensive journey. Sending a copy of every title everywhere can strand capital in quiet shelves. Placement is the choice between those costs, constrained by the service being offered.
Public sources and evidence boundary
- MIT OpenCourseWare: Stochastic Inventory Models and Supply Contracts (2009)
- MIT OpenCourseWare: Inventory Management (2016)
Public material checked on 18 October 2026. Sources support the named definitions or frameworks; the scenarios, diagrams, calculations and proposed decisions are original teaching examples. No private company systems or performance data are described.