Quiz 2 Review — Chapters 18 & 19

Forecasting & S&OP
in 15 Minutes

Click each question to reveal the answer. Build your intuition from "why" → "how" → "so what."

0 / 0 revealed
1

Forecasting — The "Why" & "What"

Start here. Every supply chain decision begins with a prediction.

1
Why can't companies just wait and react when demand shows up?
Because every decision has a lead time. Building a new factory can take a year+; even hiring a barista takes weeks. Forecasts let companies plan ahead so capacity, materials, and labor are ready when customers arrive.
🎓 Campus Example
Your campus dining hall can't wait until 11:55 AM to start cooking lunch. They forecast how many students will show up based on the class schedule, day of the week, and whether it's midterm season.
2
"The only thing certain about a forecast is that it will be wrong." So why bother?
A wrong estimate is still far better than no estimate. The goal isn't perfection — it's reducing uncertainty enough to make good decisions. Think of it like checking the weather: the forecast isn't always right, but you'd rather have it than walk out blind.
💼 Interview tip: Interviewers love to hear "all forecasts are wrong, but some are useful." Show you understand the goal is to minimize error and plan for uncertainty, not to achieve perfection.
3
What are the four basic types of forecasting, and when would you use each?
1. Qualitative — Expert judgment when you have no historical data (e.g., launching a brand-new product). Includes Delphi method and market research.

2. Time Series — Uses past demand patterns to predict the future. Includes moving averages, exponential smoothing, and trend/seasonality models.

3. Causal (Regression) — Demand is explained by another variable. The X-axis is something other than time (e.g., rainfall → umbrella sales).

4. Simulation — Computer models that run many "what if" scenarios.
🎓 How to Remember
New product, no data? → Qualitative (ask experts).
Have months of sales history? → Time series (look at patterns).
Think another variable drives demand? → Causal/regression.
Complex system with many uncertainties? → Simulation.
4
Simple Moving Average vs. Weighted Moving Average vs. Exponential Smoothing — what's the real difference?
All three are averaging methods for stable demand (no strong trend). The difference is how they weight past data:

Simple MA: Every period in the window gets equal weight. Average the last N periods.
Weighted MA: You choose unequal weights (must sum to 1.0). Recent periods usually get more.
Exponential Smoothing: Uses all past data but weights decay exponentially. Only needs last forecast + last actual + α.
🎓 Concert Tickets Analogy
Predicting demand for a spring concert at your school:
SMA: Average of last 3 years' ticket sales — equal weight to each year.
WMA: Give last year 50%, two years ago 30%, three years ago 20% — because your school's popularity is growing.
ES: Start with last year's forecast + adjust by how far off you were, controlled by α.

F_t = F_(t-1) + α (A_(t-1) − F_(t-1))

Key insight: Higher α (closer to 1.0) → reacts faster to recent changes. Lower α (closer to 0) → smoother, slower to react.
5
Which forecasting method is the most widely used in business, and why?
Exponential smoothing. Six reasons make it the champion: it's surprisingly accurate, easy to formulate, easy to understand, requires little computation, needs minimal storage, and accuracy tests are simple to compute.
💼 Interview tip: If asked "pick one method," say exponential smoothing and rattle off 2–3 of those reasons. It shows you know the practical tradeoffs, not just the math.
6
How do you know if your forecast is any good? (MAD, MAPE, Tracking Signal)
Three key measures:

MAD (Mean Absolute Deviation) — Average size of your errors, ignoring direction. Ideally zero. Bigger = less accurate.
MAD = Σ|Actual − Forecast| / n

MAPE (Mean Absolute Percent Error) — Expresses MAD as a percentage of demand. Lets you compare accuracy across products with very different volumes.
MAPE = MAD / Average Demand

Tracking Signal (TS) — Detects bias (are you consistently over- or under-forecasting?)
TS = RSFE / MAD
Where RSFE = Running Sum of Forecast Errors (keeps the + and − signs).

Rule of thumb: If TS exceeds ±3 to 4 MADs, your model has a problem — probably missing a trend.
🎓 Grade Analogy
MAD = How far off your exam grade predictions are on average (e.g., off by 8 points).
MAPE = That error as a % of total points (e.g., 8%).
Tracking Signal = Are you always overestimating your grades? That's bias — time to recalibrate.
2

S&OP — Turning Forecasts into Action

The forecast tells you what customers want. S&OP decides how to deliver it.

7
What is Sales & Operations Planning, and where does it sit in the planning horizon?
S&OP (also called aggregate planning) translates business plans into broad labor and output plans for the intermediate term (3–18 months), typically in monthly buckets. It works at the product family level, not individual SKUs.

The goal: align sales expectations with operations capacity so the company can provide good customer service, control inventory, stabilize production rates, and keep costs down.
🎓 Campus Example
Think of your university's plan for next semester: how many sections of each course to offer, how many TAs to hire, how many classrooms to book. That's aggregate planning — they're not scheduling individual class times yet, just making sure the big-picture capacity matches expected enrollment.
8
What are the three pure production planning strategies? When would you pick each?
1. Chase Strategy — Match production to demand by hiring/firing workers each period. Best when labor is flexible and easy to train (e.g., seasonal agriculture, retail holiday staff).

2. Level Strategy — Keep production constant; absorb demand swings through inventory, backlogs, or lost sales. Best when hiring/firing is expensive or workforce skill matters.

3. Stable Workforce – Variable Hours — Keep the same people but flex overtime/undertime. A middle ground when you want to keep trained employees but demand fluctuates.

A mixed strategy combines two or more of these — and that's what most real companies do.
🎓 Staffing Example
Chase: A campus coffee shop hires 10 extra baristas for finals week, lays them off after.
Level: The shop keeps the same 5 baristas year-round and makes extra drinks ahead of time (batch cold brew).
Variable Hours: Same 5 baristas, but they work overtime during finals and lighter hours over break.
9
What are the four cost categories in an aggregate plan?
1. Basic production costs — Materials and direct labor.
2. Costs of changing the production rate — Hiring, training, layoffs, overtime premiums.
3. Inventory holding costs — Storage, insurance, obsolescence, capital tied up.
4. Backordering costs — Expediting, lost sales, and customer goodwill. This is the hardest to measure because lost goodwill is intangible.
💼 Interview tip: When asked "which S&OP cost is hardest to measure?" the answer is backorder/lost-sales cost. Explain that you can estimate holding costs and overtime, but the dollar value of a customer who quietly leaves is nearly impossible to pin down.
10
What is yield management, and where does it work best?
Yield management (also called revenue management) adjusts price to influence demand, rather than just adjusting supply. It works best when:

✓ Fixed costs are high, variable costs are low
✓ Inventory is perishable (empty seats, unsold hotel rooms)
✓ Demand is segmentable (business vs. leisure travelers)
✓ Product can be sold in advance
🎓 Concert / Airline Example
A concert at your school arena has 5,000 seats. Once the show starts, unsold seats are worth $0 — perishable inventory. So they might offer early-bird pricing at $30, regular at $50, and last-minute/floor seats at $80. Same product, different prices, different customer segments. Airlines do exactly the same thing with every flight.
3

Connecting the Dots

See how forecasting feeds directly into S&OP, and S&OP feeds into everything else.

11
How does forecasting flow into S&OP, and then into MRP?
It's a top-down cascade:

Forecast (Ch 18) → "We expect to sell ~10,000 mattresses next quarter."
S&OP / Aggregate Plan (Ch 19) → "We'll produce 900 in Month 1, 950 in Month 2 across all mattress types."
Master Production Schedule → Disaggregate into specific models per week.
MRP (Ch 21) → Explode the BOM to figure out exactly what parts to order and when.

Key idea: A bad forecast cascades into a bad aggregate plan, a bad MPS, and a bad MRP — garbage in, garbage out.
12
If your tracking signal keeps drifting positive, what's happening and what do you do?
A consistently positive RSFE means actual demand keeps coming in higher than your forecasts — you're systematically under-forecasting. This signals your model is missing an upward trend.

Fix: Switch to (or enable) exponential smoothing with trend, which adds a trend-adjustment component (the extra parameter T). Or increase your α to react faster, or add trend via regression.
🎓 Real-World Clue
If every month your campus bookstore sells more hoodies than predicted, and the tracking signal crosses +4 MADs, it's not random — maybe the school's social media popularity is growing. Time to update the model to include that upward trend.
✓

Check Your Understanding

5 quick questions — tap to answer, get instant feedback.

1. Which forecasting method includes ALL past data in its calculation but weights recent data more heavily?
Simple Moving Average
Weighted Moving Average
Exponential Smoothing
Linear Regression
Exponential smoothing is the only method that uses ALL past data (through the recursive formula), while weighting recent data more heavily via α.
2. A manager wants to match production exactly to demand each month by adjusting workforce size. This is a:
Level strategy
Chase strategy
Mixed strategy
Yield management
Chase strategy = "chase" demand by hiring and laying off. The workforce flexes to match the production rate to demand each period.
3. Tracking Signal = RSFE ÷ ___?
MAD
MAPE
Standard Deviation
Average Demand
TS = RSFE / MAD. It tells you whether errors are accumulating in one direction (bias). Values beyond ±3–4 indicate a problem.
4. Which S&OP cost is typically the hardest to accurately measure?
Basic production cost
Inventory holding cost
Backordering cost
Overtime cost
Backorder costs include lost goodwill — when a frustrated customer silently switches to a competitor, that cost is real but nearly impossible to quantify.
5. Yield management works best when inventory is perishable, fixed costs are high, and demand is:
Perfectly predictable
Segmentable by customer type
Completely random
Constant year-round
Yield management requires segmentable demand so you can charge different prices to different customer groups (e.g., business vs. leisure travelers, early-bird vs. last-minute buyers).
YOUR SCORE
0/5