Strategic Capacity Week 1-3

🕰 The "Horizon" vs. The "Period" Back of Card:

• Period: The bucket of time (Day, Week, Month).

• Horizon: How far into the future you are looking (Next week? Or 5 years from now?).

🔮 What is a Demand Forecast?

An estimate of what customers will want in a future time period.

: 📉 The Golden Rule of Accuracy

"Forecasts are always inaccurate."

: 🍔 The "Hangry" Customer Scenario (McDonald's) Scenario: You run a McDonald's. Why do you need to forecast customer traffic?

Quality Management.

• In the service industry, Quality = "Having the product when the customer demands it."

• If you don't forecast the lunch rush, you run out of fries = Bad Qualit

🏭 The "Big Rigs" Decision (Capacity) Scenario: You need to decide if you should build a massive new factory for a new videophone. Is this a Short-Range or Long-Range forecast?

Long-Range (Strategic). Time: 2+ years.

• Why: You are planning facilities, technology, and supply chain design

📅 The "Shift Manager" Decision Scenario: You need to know how many call centre operators to schedule for next Tuesday

Short-Range (Operational).

• Time: Days, weeks, or months (up to 2 years).

• Why: You are planning purchasing, job scheduling, and inventory level

: 🎨 The Home Depot Paint Mystery The Problem: The Paint Factory is making white paint. Home Depot is planning a huge sale on white paint but doesn't tell the factory. What happens?

A Supply/Demand Mismatch.

• The Term: Lack of Collaboration.

• Result: Home Depot runs out of paint (Stockout) because the factory didn't know they needed to produce extra

The Coca-Cola Solution The Fix: How does Coca-Cola stop the "Paint Mystery" from happening with their bottlers?

Collaborative Forecasting.

• Coca-Cola shares their promotion plans with the bottlers.

• The bottlers update their production capacity before the promotion starts.

🎲 Betting on the Future (Time) Question: Which forecast is safer to bet money on: What sales will be next week, or what sales will be 5 years from now?

Next Week (Short-term).

• The Rule: Long-term forecasts are less accurate than short-term forecasts because there is more uncertainty

Betting on the Product (Aggregation) Question: Which forecast is more accurate: Total sales for all Toyota cars, or sales for just the Red Toyota Camry?

All Toyota Cars (Aggregate). The Rule: Aggregate forecasts (groups of products) are more accurate than disaggregate forecasts (individual items). Errors tend to cancel each other out in big group

The "Telephone Game" Effect Question: Who has the worst forecast data: The store manager selling to the customer, or the raw material supplier 5 steps away?

The Supplier (Farther away) The Rule: The farther up the supply chain you are from the consumer, the greater the distortion of information

: 📈 The 3 "Vibes" of Demand (Patterns) Hint: How does the line move on the graph?

. Trend: Gradual long-term movement (Up like TikTok, or down like cable TV).

2. Cycle: The Rollercoaster. Up and down over years (The Economy/Recessions).

3. Seasonal: The Heartbeat. Repetitive short-term spikes (Turkeys in Oct, Sunscreen in July

🔮 Qualitative vs. Quantitative (The showdown) Hint: Do you have a calculator or a crystal ball?

Qualitative: NO historical data. Uses intuition & experts (e.g., New VR Headsets).

• Quantitative: HAS historical data. Uses math & statistics (e.g., Selling milk or paper towels

📜 The "First Step" Rule Hint: Before you calculate anything, what must you do?

Understand the objective.Why: You need to know what decision this forecast supports (e.g., "Do I build a factory?" vs. "How much lettuce do I buy?")

🤥 The "Pinocchio" Data Trap (Sales vs. Demand) True or False: "My sales report shows exactly what customers wanted."

: FALSE.

• Sales: What you actually sold.

• Demand: What customers wanted to buy.

• The Trap: If you ran out of stock (Stockout), your sales are zero, but demand was high. If you ignore this, your forecast will fail

🍜 The "Chicken Noodle Soup" Anomaly Scenario: You usually sell zero soup in July. Suddenly, you sell 1,000 cans. Why?

Internal Actions (Promotions/Pricing).

• The Lesson: Demand isn't always magic; sometimes you caused it by lowering the price. You must adjust your forecast for promotions.

The Delphi Method (Qualitative) Hint: How do you get experts to agree without a fistfight?

Iterative Questionnaires. Send surveys to experts → Summarize results → Send them back again → Repeat until everyone agrees (Consensus

🎮 Job: Launching the PS6 (New Console) Situation: Sony is launching a brand new console. No sales history exists. Which method do you use?

Qualitative Methods. Tools: Executive Opinion, Consumer Surveys, or Delphi Method.

• Why: You cannot use math (Time Series) because "the past does not exist" for this product

Job: Restocking 2% Milk Situation: You run a grocery store. Milk sales are stable and boring every week. Which method do you use

Quantitative (Time Series).

• Tools: Moving Average.

• Why: The situation is stable and you have years of data. "Past demand is a good indicator of future demand".

Case Study: The "Stable" Enrollment Trick The Clue: State University says enrollment (Sales) is flat/stable. Is everything okay? Back of Card:

NO. Demand is dropping. • The Analysis: They kept "Sales" up by lowering their standards and accepting less qualified students.

• The Hack: Always look for what the organization did to force the numbers to look good. Their true demand (applications) is crashing

📉 Case Study: The "Tuition" Factor The Clue: The University raised tuition every year. How does this affect your forecast model?

It requires a Causal Model.

• The Analysis: You can't just look at past years (Time Series). You must look at the relationship between Price (Tuition) and Demand (Applications). As Price goes up, Demand likely goes down

🕵 Case Study: Determining "Quality" The Clue: Parents are questioning the "value" of the degree. How do you forecast "Value"

Use Consumer Surveys (Market Research).

• The Analysis: There is no historical number for "reputation." You must go out and ask the customers (parents/students) directly to get this dat

The "Rearview Mirror" Method Question: What is the main assumption of Time Series forecasting?

The past predicts the future."

• The Logic: It assumes that history repeats itself. It relates forecast to only one factor: Time.

• Warning: It ignores outside factors like competitor pricing or the economy

The "Smoothie" (Moving Average) Question: Why do we use a Moving Average instead of just looking at last month's sales?

To smooth out the "noise."

• The Logic: It dampens random spikes and dips to show you the general level of demand.

• The Vibe: Stable and calm, but slow to react

The "Alpha" Knob (α) Question: In Exponential Smoothing, what is α?

The "Reaction Speed" Knob.

• The Logic: It is a weighting factor (between 0 and 1).

• High α (0.8): Reacts fast to recent changes (Panicky).

• Low α (0.1): Reacts slow, ignores noise (Chill

The "Lag" Problem Scenario: Sales are skyrocketing every month (Trend). You are using a simple 3-Month Moving Average. Will your forecast be too high, too low, or correct?

Too Low (Under-forecasting).

• The Logic: Moving averages "lag" behind trends because they are anchored by old data. You will run out of stock.

• The Fix: You need a Linear Trend Line or Adjusted Exponential Smoothing,.

Weighted vs. Simple Moving Average Question: Why would you use a Weighted Moving Average instead of a Simple one?

To prioritize the "Now."

• The Logic: It lets you give more weight (importance) to the most recent month.

• Example: If you think last month matters more than 3 months ago, use Weighted,

The "Ruler" Method (Linear Regression) Question: When should you use the formula y=a+bx? Back of Card:

When there is a clear, straight-line TREND.

• The Logic: This draws a straight line through your data.

• The Catch: It assumes the future will follow that exact straight line forever. It cannot handle sudden changes in direction,.

💻 Job: The HiTek Technician (Adjusted Smoothing) Scenario: HiTek's demand is growing steadily. Simple Exponential Smoothing (α) keeps missing the mark. What extra ingredient do you add?

The Trend Smoothing Constant (β).

• The Logic: This is Adjusted Exponential Smoothing.

• α (Alpha): Smooths the randomness.

• β (Beta): Tracks the trend.

• Result: The forecast "catches up" to the growth

Job: Wishbone Farms (Seasonality) Scenario: You sell 80% of your turkeys in Q4. If you just average the whole year, you will fail. What do you calculate first?

The Seasonal Factor.

• The Formula: (Demand for that Quarter) ÷ (Total Annual Demand).

• Example: If Q4 Factor is 0.37, it means 37% of your business happens in Q4. You multiply your annual forecast by 0.37 to get the Q4 target,.

📉 Diagnostic: The "Flat Line" Error Scenario: Your forecast is a straight horizontal line, but reality is a squiggly line. What method are you likely using?

A Moving Average (on volatile data).

• Analysis: Moving averages smooth out everything, including useful patterns like seasonality. If your business is seasonal (like turkeys), a Moving Average destroys the useful data. You need Seasonal Adjustments

🧠 The "Common Sense" Override Question: The math says sales will double next month because of a trend line. But you know a competitor just opened across the street. What do you do?

Switch to Causal or Qualitative Methods.

• Analysis: Time Series only looks at Time. It is blind to competitors, prices, and promotions.

• The Rule: Math is useful, but it is not a crystal ball. When the environment changes, the math break