1. The story: two companies that stopped guessing each other's needs
In the 1980s, Procter & Gamble and Wal-Mart worked out a closer partnership, starting, in the Harvard case's telling, with a canoe trip between P&G sales executive Lou Pritchett and Sam Walton. [Sourced: HBS case 907-011 abstract] Wal-Mart also built the machinery: a distribution center in 1970 and, by 1975, IBM systems linking store and DC inventory. [Sourced: SCDigest timeline] By 1997 it gave suppliers daily store-level sales through Retail Link. [Sourced: Talk Business 1997]
Across the Atlantic, Barilla had the opposite problem. In the late 1980s its distributors' orders swung wildly. Its proposal: Barilla, not the distributors, would decide shipment quantities, a continuous replenishment program. Some internal teams and distributors resisted. [Sourced: HBS Barilla (A) abstract]
Lesson in one line: the hard part of replenishment is rarely the math. It is who holds the information and who gets to decide.
2. The one idea
Every item has two questions forever: when do I order, and how much? Replenishment is the set of rules, people and software that answer them for every SKU at every location. [General]
3. The kitchen-table version
Your home pantry. You keep a stock of water bottles. You don't wait for zero because the store trip takes time. You reorder when you hit a line, say 8 left, because the trip and the use during it are covered. You buy a box rather than one because trips cost effort. That is a reorder point and an order quantity. [General]
4. The five building blocks
- Reorder point (ROP) = demand during lead time + safety stock. When stock position falls to ROP, order. [General]
- Safety stock: the cushion for surprises in demand or supply (see Module 9).
- Order quantity: how much. Classic answer is EOQ.
- Review style: check continuously, or on a schedule (e.g. weekly).
- Min/max: simplest practical form. Order up to max when stock falls to min. Common in stores and warehouses, because staff can follow it without software. [General]
EOQ in one paragraph
Ordering often costs money per order; holding stock costs money per unit per year. EOQ is the order size that balances the two: EOQ = square root of (2 x annual demand x cost per order / holding cost per unit per year). Ford W. Harris published it in 1913 in Factory magazine. A 1990 Operations Research paper notes it was then little noticed for decades and rediscovered in 1988. [Sourced: Harris 1913 reprint; INFORMS paper] The model assumes steady demand and fixed costs, so treat it as a starting point. [General]
5. Worked example (the $12 water bottle) [Illustrative, computed by script]
- Demand 20 a day (7,300 a year); lead time 7 days; safety stock 30.
- ROP = 20 x 7 + 30 = 170. Order when stock position hits 170.
- Cost to place an order $50; holding cost = 25% of $5 unit cost = $1.25 a year.
- EOQ = sqrt(2 x 7,300 x 50 / 1.25) = about 764 bottles, roughly 9.6 orders a year. Order cost about $478 a year, holding cost on average stock (half of 764) about $478. The two roughly match, which is the point of EOQ.
- Min/max version for a shelf: min 170, max 934 (ROP + EOQ).
6. Who decides: four models [General]
| Model | Who decides | Typical fit |
|---|---|---|
| Retailer-managed | Buyer or system at retailer | Most retail |
| Vendor-managed inventory (VMI) | Supplier, using the retailer's stock and sales data | Strong partner, shared data |
| Continuous / auto replenishment | Rules trigger orders from sales or warehouse withdrawals | High-volume staples |
| CPFR (Module 5) | Joint plan and exceptions | Big promotions, key accounts |
Evidence: grocery's 1990s Efficient Consumer Response work spread continuous replenishment; a 1995 trade report says most programs were then vendor-managed and based on retailer warehouse withdrawal data. [Sourced: Supermarket News 1994, 1995] Campbell Soup's program was studied in an academic paper. [Sourced: Production and Operations Management, 1997, title/abstract only]
7. Store vs DC
- DC replenishment pulls from suppliers in bigger lots with long lead times.
- Store replenishment moves small quantities from DC, often daily, limited by shelf space, pack size and delivery calendars.
- Push vs pull: new or scarce items are usually pushed by a central plan; established items are pulled by sales. Vendors describe blending both. [Sourced: Blue Yonder page; vendor claim]
- Rule of thumb: the closer to the customer, the smaller the lot and the shorter the horizon. [Our view]
8. Where AI fits
- Better demand input per SKU and store, daily, even for new items (vendor claims; test before believing). [Sourced: Pattern, o9 pages, claims]
- Learning lead times and their variability instead of fixed numbers. [General]
- Setting safety stock by service goal and cost, SKU by SKU. [General]
- Auto-generating orders and flagging exceptions for people. [General]
- Gartner predicts only 5% of organizations will make at least 10% of supply chain planning decisions autonomously by 2030, so expect assisted, not hands-off. [Sourced: Gartner, Sept 24 2026]
- Forward thread: an agent that reorders for you and negotiates with a supplier's agent is where this heads. [Our view]
9. Where it breaks
Wrong stock records (system says 12, shelf has 0), unreliable lead times, minimum order quantities, case packs, promotions, substitutions. Bullwhip (Module 4): each tier reordering from the one below amplifies swings. [General; Sourced: MIT Sloan on bullwhip]
10. Exercise
Your bottle sells 12 a day, lead time 5 days, safety stock 15. Find ROP. If lead time becomes 9 days, what happens? Answer: 75; 123.
11. Quiz
- Two parts of a reorder point? Lead-time demand and safety stock.
- What does EOQ balance? Ordering cost and holding cost.
- What did the Barilla proposal change? Who decides shipment quantities.
- Why will AI assist rather than replace planners soon? Gartner's 5% by 2030 prediction; messy data.
Sources
- HBS Pritchett case: https://www.hbs.edu/faculty/Pages/item.aspx?num=34010
- HBS Barilla (A): https://www.hbs.edu/faculty/Pages/item.aspx?num=22411
- SCDigest: https://www.scdigest.com/ASSETS/ON_TARGET/12-07-27-1.php
- Talk Business 1997: https://talkbusiness.net/1997/06/internet-provides-link-between-wal-mart-vendors/
- Harris 1913: http://userhome.brooklyn.cuny.edu/irudowsky/CIS10.31/articles/EOQModel-OriginalPaper.pdf
- INFORMS: https://pubsonline.informs.org/doi/10.1287/opre.38.6.937
- Supermarket News 1994: https://www.supermarketnews.com/grocery-operations/ecr-begins-taking-shape ; 1995: https://www.supermarketnews.com/grocery-operations/change-in-continuity-
- Campbell: https://ideas.repec.org/a/bla/popmgt/v6y1997i3p266-276.html
- Blue Yonder: https://info.blueyonder.com/retail-planning-category-management/what-is-blue-yonder-allocation-replenishment
- Gartner: https://www.gartner.com/en/newsroom/press-releases/2026-09-24-gartner-predicts-only-5-percent-of-organizations-will-make-at-least-10-percent-of-supply-chain-planning-decisions-autonomously-by-2030