Inventory Is a Way of Buying Time
Demand, lead time, reorder points, safety stock, service levels and the bullwhip effect.
Part 3 of 18 · Before the Shelf
Maya cannot print yesterday's missing book
Sathya can decide to buy a book today. Maya cannot respond by printing it yesterday. If eComm Inc sells an average of 100 copies a day and fresh stock takes ten days to become usable, the replenishment decision must be made while books are still on the shelf.
A forecast describes possible demand. An order commits to supply. The two are related, but neither can substitute for the other. Selling 80 copies before a stockout also does not prove that demand was only 80; unavailable goods hide some demand.
A small model, with its assumptions visible
In a simple continuous-review model:
Reorder point = average demand per day × usable replenishment days + safety stock.
With 100 books per day, ten days and an explicitly chosen 300-book buffer, the threshold is 1,300 books. The buffer is a teaching assumption, not a calculated service guarantee. MIT's inventory course provides the general background; the numbers and decision below are our own.
Compare the threshold with inventory position, defined here as usable on-hand stock plus relevant open supply minus backorders. If 600 books are on hand, 500 are due, and 100 are backordered, position is 1,000. Under this policy, replenishment is triggered. The 500 due must be credible supply within the relevant planning horizon; an overdue promise cannot protect today's shelf merely because it remains open in a system. If a system instead uses stock already net of reservations, it must not subtract those commitments a second time.
A receiving delay can look like a buying shortage
If congestion adds four days, expected demand during replenishment rises from 1,000 to 1,400. Keeping the same assumed 300-book buffer raises the threshold from 1,300 to 1,700. The extra 400 books compensate for a slower process. They do not demonstrate that customers became less predictable.
This example holds daily demand and the chosen buffer constant. In practice, a longer or more variable lead time may also change the buffer. Conversely, fixing the receiving delay can reduce the required pipeline without asking customers to accept worse service.
Measure the whole interval. A supplier can improve production by one day while the destination adds two days of waiting. A transport-only chart would celebrate improvement while the shelf becomes harder to replenish.
What does a service level mean?
A cycle service level asks how often a replenishment cycle avoids a stockout under a defined model. A fill rate asks what share of demand is supplied from the intended stock. They are different denominators. Nine successful cycles out of ten does not establish that 90% of units were filled.
Demand variability, supplier reliability, substitutes, shelf life and shortage consequences all matter. A school workbook needed by a fixed date differs from a slow-selling reference book. More safety stock is not a universal answer: it may become obsolete or occupy space needed by faster-moving goods.
Review frequency matters too. A buyer who checks once a week must protect the period until the next review as well as replenishment time. A continuous-review reorder point should not be pasted unchanged into a weekly ordering process.
Why a modest change can become a large upstream order
Imagine customers buy 10% more books. eComm Inc orders an extra batch to protect itself. Maya adds another batch because paper is bought in large lots. The paper supplier sees a sharp jump in orders and interprets it as permanent demand. Each local decision can amplify the original signal.
The classic bullwhip literature identifies demand updating and order batching among the mechanisms. Our fictional example illustrates the idea; it is not a measured amplification factor. Sharing actual demand, distinguishing promotions from lasting growth and making batch rules visible can help diagnose it. No single software tool removes the underlying incentives.
Find out what the extra stock is protecting
Before raising the buffer, eComm Inc should locate the delay. If four extra days sit between arrival and usable release, more purchasing treats the symptom while the receiving queue keeps consuming capital. If demand has become more variable, however, restoring the old dock speed may leave a genuine need for protection.
Maya also needs to know what the forecast means. A promotion estimate, a forecast range and an accepted purchase order support different production decisions. Treating all three as firm demand can pass a temporary spike back to the paper supplier and eventually into an unnecessary machine purchase from Mayan.
The practical test is simple: the truck arrives two days earlier, but stock becomes usable on the same day as before. Transport improved. The buyer's replenishment clock did not. Measure the interval the inventory must actually bridge before deciding how much inventory to hold.
Public sources and evidence boundary
- MIT OpenCourseWare: Inventory Management (2016)
- Lee, Padmanabhan and Whang: The Bullwhip Effect in Supply Chains (1997), public abstract
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.