Exam 1-Supply Chain

Operations management (OM)

How an organization turns inputs (people, materials, money, information) into outputs (goods and services) customers value, as efficiently and reliably as possible. Looks inside one firm.

Supply chain management (SCM)

Planning and coordinating sourcing, production, logistics, and the flows of information and money across the companies that bring a product to the customer.

Supply chain

The network of suppliers, manufacturers, distributors, carriers, and retailers that gets a product to the end customer.

Upstream

Toward the suppliers.

Downstream

Toward the customer.

Three flows in a supply chain

Materials/products flow downstream; money flows upstream; information (orders, forecasts) flows both ways.

Which direction does a purchase order flow?

Upstream. Product shipments and warranty replacements flow downstream.

Operations strategy

The set of choices a firm makes about how it designs and runs its operations (products, processes, technology, capacity, people, inventory, facilities, suppliers) to support how it competes.

Value proposition

The constellation of value dimensions that matter to a customer; essentially a promise made to the customer. Includes order qualifiers and an order winner.

Order qualifier

The minimum level of value required for a firm to even be considered. Gets you considered.

Order winner

The dimension that differentiates one firm from another and makes the customer choose you. Gets you chosen.

Car rental example (qualifier vs. winner)

Airport availability is the order qualifier; lowest price is the order winner.

Operational capabilities

Unique, superior abilities based on a firm's routines, skills, and processes that meet customer expectations and are difficult to imitate. Must be intentionally designed.

Four operational capabilities

Cost, Quality, Time, Flexibility.

Cost capability

Low cost; efficient processes (e.g., Costco, IKEA).

Top quality

Outstanding product/service, often with high customer contact (e.g., Rolex, Ferrari).

Consistent quality

Meets specs every time; processes monitored to prevent defects (e.g., McDonald's, Snickers).

Delivery speed

Quickly filling an order; reducing lead time (e.g., Netflix, UberEats).

On-time delivery

Meeting delivery-time promises (e.g., UPS, Alaska Airlines).

Development speed

Quickly introducing new products (e.g., Zara, H&M).

Customization

Meeting each customer's unique needs; low volume, close contact (e.g., Ritz-Carlton).

Variety

Handling a wide assortment efficiently (e.g., Amazon).

Volume flexibility

Quickly ramping production up or down; needs excess capacity or inventory (e.g., USPS).

Three pillars of an effective operations strategy

1) Externally consistent (delivers the value proposition). 2) Internally consistent (capabilities reinforce each other). 3) Dynamic (evolves over time to maintain or extend advantage).

GM vs. Toyota lesson (2004-2007)

GM was #1 in sales but lost $9.7B; Toyota was #1 in profit (+$36.3B). Sales volume is not the same as a strategy that fits.

Walmart/Target lesson

You can't be good at everything. Walmart wins on price, selection, and in-stock; Target beat it on cleanliness, quality, and checkout. Trade-offs are intentional.

McDonald's order qualifiers

Low price and consistency.

McDonald's order winner

Speed (fast service).

McDonald's incrementalism

Menu grew from about 10 to roughly 150 items and operations broke down. Adding variety eroded the capabilities (speed, consistency) that created the advantage.

Efficient supply chain

Goal: low cost (minimize inventory, maximize process efficiency). Demand: highly predictable. Product life cycle: long. Contribution margin: low. Products: stable, basic (household staples).

Responsive supply chain

Goal: flexibility (react quickly to changing demand). Demand: unpredictable. Product life cycle: short. Contribution margin: high. Products: changing, innovative (fashion).

Push system

Production and distribution driven by forecasts, made before demand is known. Fits long lead times, low demand uncertainty, high economies of scale (e.g., toothpaste, cereal, paper towels).

Pull system

Triggered by actual customer demand; no finished-goods inventory, responds to specific orders. Fits short lead times and high demand uncertainty (e.g., Dell build-to-order, custom furniture).

Push-pull (hybrid) system

Early stages run on push (forecasts), late stages run on pull (actual demand).

Push-pull boundary

The point where activities stop being forecast-driven (push) and start being order-driven (pull).

Postponement

Delaying an irreversible, product-defining decision until better demand information is available.

Moving the push-pull boundary upstream

More decisions are postponed until demand is known (more pull). Moving it downstream means more is committed on forecasts.

Efficient/responsive vs. push/pull (key distinction)

Efficient/responsive describe strategic priorities. Push/pull describe how activities are triggered. They are different things.

Zara supply chain strategy

Strategic priority: responsiveness. Delivers a new coat from signal to Fifth Avenue in about 25 days. Postpones commitment rather than forecasting perfectly.

Zara manufacturing split

Trendy items (about 55%) made nearby in small batches with short lead times. Basics (about 45%) made farther away at lower cost with longer lead times.

Zara push activities

Secure fabric and raw materials, plan initial capacity, develop possible styles, produce some basics.

Zara pull activities

Final style and quantity decisions, in-season production, store allocation, replenishment, color/feature modifications, all triggered by market data.

Inventory

A stock of materials used to satisfy customer demand or support production of goods or services.

Inventory management

Planning and controlling inventories to meet the organization's competitive priorities.

Three types of inventory

Raw materials/components (unprocessed purchased inputs), work-in-process/WIP (partially processed), finished goods (complete and ready to sell).

Four purposes of inventory

1) Buffer against uncertainty. 2) More efficient production. 3) Lower product cost via quantity discounts. 4) Shorter delivery times via geographic coverage.

Newsvendor model

Model for one-shot ordering under uncertain demand with a known distribution, known profit/loss per unit, and a goal of maximizing expected profit. Fits perishables and short selling seasons (newspapers, Christmas trees, one-night concert programs). Does NOT fit steady year-round demand or weekly restocking.

Cu (cost of understocking)

Price - Cost (+ goodwill cost per lost sale, if given). The profit lost per unit you were short.

Co (cost of overstocking)

Cost - Salvage value. A disposal fee makes salvage negative, so it adds to Co.

Critical ratio (CR)

Cu / (Cu + Co). The target service level, i.e., the probability that demand is less than or equal to Q*. It is a probability, not a quantity.

Optimal order quantity Q* (uniform demand)

Q* = min + CR x (max - min)

If Cu > Co

CR > 0.5, so order above the middle of demand.

If Co > Cu

CR < 0.5, so order below the middle of demand.

If Cu = Co

CR = 0.5, so order the midpoint of demand.

Newsvendor profit (when asked)

Revenue from units sold + salvage on leftovers - cost of everything ordered - goodwill on lost sales.

Bullwhip effect

Small fluctuations in consumer demand get amplified as orders move upstream through the supply chain. Upstream order variability exceeds consumer-demand variability.

Why the bullwhip happens

Each tier sees only the orders from the tier directly below it. Only the retailer sees real consumer demand.

Beer chain structure

Consumer, Retailer, Wholesaler, Distributor, Manufacturer. Orders flow upstream, beer flows downstream.

Causes of the bullwhip effect

Order batching, price promotions/forward buying, shortage gaming, demand-forecast updating at every stage, and longer lead times.

Remedies for the bullwhip effect

Share actual point-of-sale (POS) demand data, shorten lead times, order smaller and more often, stabilize prices (everyday low pricing), avoid shortage gaming, coordinate forecasting and replenishment, vendor-managed inventory (VMI).

Consequences of the bullwhip effect

Excess inventory (holding and opportunity costs), erratic production, stockouts (lost sales and market share), markdowns.

Beer Game costs

Holding cost $0.50/unit/week; stockout cost $1/unit/week (stockouts cost 2x holding).

Bullwhip example: P&G diapers

Baby population is stable, yet retailer orders swung and raw-material orders swung even more. Causes: promotions, batch ordering, fear of shortages.

Bullwhip example: Cisco

Customers double- and triple-ordered scarce routers; orders then cancelled. Result: $2.2B inventory write-off in a quarter, 8,500 layoffs.

Analytics and optimization (AI toolkit)

People write the rules and the goal; the computer searches for the best answer. Nothing is learned from data. Example: UPS ORION routing.

Conventional machine learning

Learns patterns from historical data to predict, classify, recommend, or decide. Output is a label, score, or number. Example: Amazon demand forecasts.

Generative AI

A branch of ML that learns the structure of its training data and generates new content (text, images, code). Example: Walmart supplier-negotiation chatbot.

Relationship between GenAI and ML

GenAI is a kind of ML, not a separate thing. The difference is the task and output, not whether it predicts.

Hallucination

Output from a generative model that is fluent and confident but factually false or made up.

Where AI helps the supply chain

1) Demand forecasting (conventional ML, the most common and proven use). 2) Real-time visibility and tracking. 3) Insight to action via agentic AI.

Agentic AI

An agent that can detect a supplier delay, find alternative suppliers, recalculate order quantities, re-route deliveries, and test the plan in a digital twin.

Four AI failure modes

1) Bad/dirty data (garbage in, garbage out). 2) Hallucinated trends (invented demand). 3) Regime change (the past no longer predicts the future). 4) No human check.

MIT 2025 AI lesson

About 95% of 300+ corporate AI pilots showed zero measurable return. The limiting factor is rarely the model; it is data readiness, integration with old systems, and domain expertise.

Why better forecasts help

Less overstock and fewer stockouts, plus a smaller bullwhip effect (McKinsey: 20-50% lower forecast error, up to 65% fewer lost sales).