Forecasts in Supply Chain Decision-Making

Introduction

A forecast is not a decision. It is an informed estimate of future demand, supply, prices, lead times, capacity requirements, or other variables that influence action. In supply chain management, the value of forecasting lies in translating uncertain information into coordinated decisions about inventory, procurement, production, transportation, labor, capacity, revenue, and customer service.

This distinction is strategically important. A forecast may be statistically accurate yet operationally unhelpful if it is issued at the wrong level of aggregation, ignores constraints, or is not connected to decision rules. Conversely, a forecast with measurable error can support excellent decisions when uncertainty is explicitly represented, response options are flexible, and the cost of error is understood.

Supply chain forecasting therefore has two linked dimensions:

1. Prediction: estimating what is likely to occur.
2. Decision design: determining what the organization should do given the estimate, uncertainty, constraints, and consequences of action.

Forecasts are used across multiple time horizons. Strategic forecasts support network design, technology investment, and market-entry decisions. Tactical forecasts support sales and operations planning (S&OP), inventory policies, procurement contracts, and workforce plans. Operational forecasts support replenishment, scheduling, routing, and daily execution. Forecast quality must be assessed relative to the horizon, aggregation level, decision, and cost of error.

Learning Objectives

By the end of this chapter, the reader should be able to:

• distinguish forecasts from plans, targets, budgets, and scenarios;
• translate forecasts into inventory, safety-stock, capacity, procurement, workforce, logistics, revenue, and service decisions;
• calculate and interpret basic demand and lead-time uncertainty measures;
• explain the role of S&OP and integrated business planning (IBP);
• evaluate demand shaping, constrained planning, and cost-to-serve trade-offs;
• design a forecast-to-decision process with appropriate governance and performance measures;
• identify limitations, ethical risks, and appropriate uses of AI-enabled forecasting.

1. Forecasts, Plans, Targets, and Scenarios

A forecast is an estimate of a future outcome based on available information. A plan is a selected course of action intended to achieve organizational objectives. A target is a desired outcome, often established by management or a commercial function. A budget is a financially expressed plan, usually constrained by resources and accounting rules. A scenario is a coherent representation of a possible future, often used to test robustness rather than to state the single most likely outcome.

Confusing these concepts creates predictable problems. If a sales target is entered into the forecasting system as though it were an unbiased estimate of customer demand, the resulting production and inventory decisions may be systematically distorted. A target can be higher or lower than the statistical forecast for legitimate strategic reasons, but the difference should remain visible.

A useful planning architecture separates:

• Baseline forecast: the expected outcome assuming normal conditions and no discretionary intervention.
• Consensus forecast: a cross-functional view that incorporates validated commercial, market, and operational information.
• Plan: the actions selected after considering constraints, costs, policies, and strategic priorities.
• Exception assumptions: explicit events such as promotions, product launches, strikes, competitor exits, or regulatory changes.

Forecasts should also be defined by four dimensions: product or service, location or market, time bucket, and decision horizon. A monthly national forecast may be suitable for capacity planning but unsuitable for warehouse replenishment, which may require a weekly item-location forecast.

2. Forecast Error and Uncertainty

No forecast is perfectly accurate. A forecast should therefore be accompanied by an estimate of uncertainty and a statement of assumptions. Let actual demand in period t be A_t and the forecast made for that period be F_t. Forecast error is:

A positive error indicates under-forecasting; a negative error indicates over-forecasting. The sign convention must be stated because some organizations use the reverse convention.

Common accuracy measures include:

Here, n is the number of observations. MAE is expressed in the units of demand and is easy to interpret. RMSE penalizes large errors more heavily and is useful when major misses are especially costly. MAPE is scale-independent but becomes unstable or undefined when actual demand is zero or very small. For intermittent demand, weighted absolute percentage error, scaled errors, or measures based on the forecast distribution are usually preferable.

Bias measures systematic over- or under-forecasting. One simple measure is mean error:

A near-zero ME does not guarantee good forecasting: large positive and negative errors can cancel. Bias should be examined alongside dispersion and by relevant segments such as product family, customer, and region.

Forecast accuracy is not the same as forecast value. Suppose two products have equal MAE, but one is cheap and plentiful while the other is expensive, capacity-constrained, and essential to a contractual service commitment. The same statistical error can have radically different operational consequences. Decision-oriented evaluation should therefore consider weighted error, service impact, obsolescence exposure, expediting cost, lost contribution, and working capital.

Demand and Lead-Time Variability

For a replenishment item, demand during the replenishment lead time is uncertain. If average demand per period is μ_D, demand standard deviation per period is σ_D, lead time is L periods, and demand periods are independent with constant variance, the standard deviation of demand during a fixed lead time is:

This relationship assumes independent demand across periods and a known, fixed lead time. If lead time is random, with mean μ_L and standard deviation σ_L, a commonly used approximation is:

The approximation assumes demand and lead time are independent and uses a first-order variance decomposition. If they are correlated, or if demand is highly intermittent, simulation or a full empirical distribution is safer.

For a cycle-service target represented by a standard-normal quantile z, safety stock under the fixed-lead-time assumptions is:

The reorder point is:

where μ_DL = μ_D L is expected demand during lead time. Safety stock protects against uncertainty; it does not replace the expected lead-time demand.

For random demand and random lead time under the independence approximation:

These equations apply most directly to continuous-review replenishment with approximately stable demand and a specified service interpretation. A cycle-service level is the probability of avoiding a stockout in a replenishment cycle; it is not the same as fill rate, which is the proportion of demand fulfilled immediately. The distinction matters because a single stockout can involve either one unit or a large quantity.

Worked Example: Translating a Forecast into Safety Stock

A distributor sells an average of 400 units per week. Weekly demand has a standard deviation of 80 units. Supplier lead time is fixed at three weeks. Management chooses a 95% cycle-service target, for which z is approximately 1.645.

Expected lead-time demand is:

μ_D L = 400 × 3 = 1,200 units.

Demand standard deviation during lead time is:

σ_D√L = 80√3 ≈ 138.6 units.

Safety stock is:

SS = 1.645 × 138.6 ≈ 228 units.

Thus, the reorder point is approximately:

ROP = 1,200 + 228 = 1,428 units.

This result is not a universal answer. If lead time varies, demand is seasonal within the three-week window, or the target is a fill rate rather than cycle service, the policy must be recalculated. The forecast supplies the expected demand; the decision also requires uncertainty, service objectives, lead time, and replenishment mechanics.

3. Translating Forecasts into Supply Chain Decisions

Inventory and Service

Inventory decisions should distinguish cycle stock, safety stock, anticipation stock, pipeline inventory, and decoupling stock. The forecast determines expected consumption, while uncertainty and replenishment conditions determine protection requirements.

Important policy choices include:

• review frequency and order-up-to level;
• order quantity or lot size;
• target cycle service or fill rate;
• segmentation by value, criticality, variability, and predictability;
• treatment of new, obsolete, seasonal, and intermittent items.

Forecasts should be aggregated at the level where the decision is made. Forecasting an entire product family and allocating it mechanically to individual stock-keeping units can create excess inventory in slow items and shortages in fast items. Hierarchical forecasting and reconciliation help ensure that forecasts at regional, family, and item levels do not contradict one another.

Procurement

Procurement uses forecasts to determine order timing, contract volumes, supplier capacity reservations, raw-material commitments, and risk exposure. The appropriate response depends on lead time and flexibility. A long-lead, single-source component may require earlier commitment based on a probability-weighted range, whereas a short-lead commodity may be purchased later after demand becomes clearer.

Forecasts should not be converted directly into purchase orders without considering:

• existing inventory and open orders;
• minimum order quantities and economic order quantities;
• supplier capacity and reliability;
• price breaks and payment terms;
• substitution possibilities;
• cancellation and postponement rights;
• material shelf life and obsolescence.

A robust procurement plan often combines a firm commitment for near-term requirements with flexible or option-like capacity farther into the horizon.

Capacity and Production

Capacity planning translates forecast volume into required machine time, labor hours, storage space, tooling, and supplier capability. If forecast demand is D_t, the required capacity in a resource group can be expressed conceptually as:

The units must be consistent, such as machine-hours per month. Available capacity should be adjusted for maintenance, absenteeism, changeovers, yield loss, and utilization policy. A plan that matches average demand to nominal capacity may fail when variability and operational losses are included.

When demand exceeds capacity, planners can use overtime, subcontracting, alternative resources, product prioritization, postponement, allocation rules, or price-based demand shaping. The correct choice depends on contribution margin, service commitments, strategic customers, and the cost of changing capacity.

Workforce

Labor forecasts translate expected workload into staffing, shift patterns, recruitment, training, cross-utilization, and temporary labor. Workforce decisions are particularly sensitive to timing: hiring too early raises cost, while hiring too late creates overtime, quality problems, and service failures. A useful plan separates baseline staffing from flexible capacity and identifies the lead time required to train employees.

Workforce forecasts should also consider labor law, fatigue, fairness, and the risk that algorithmic systems allocate undesirable shifts disproportionately to certain groups.

Logistics and Network Decisions

Forecasts influence transportation mode, shipment consolidation, fleet requirements, warehouse space, delivery frequency, and cross-docking. A high forecast may justify planned capacity reservations, but a point forecast alone does not establish the optimal mode. The decision depends on the cost of capacity, shipment urgency, service promises, inventory position, and disruption risk.

At network level, forecasts support facility location and flow allocation. Because network decisions are difficult to reverse, scenario ranges are generally more appropriate than a single forecast. Decision-makers should test demand growth, geographic shifts, supplier disruption, fuel prices, and changes in customer service requirements.

Revenue, Pricing, and Commercial Decisions

Demand forecasts support revenue management, assortment, promotion timing, sales quotas, and pricing. However, observed demand may be censored by stockouts: sales recorded as zero may mean no demand, or simply no available inventory. Promotional forecasts must estimate incremental demand, substitution, pull-forward effects, cannibalization, and post-promotion demand depression.

A commercial forecast should therefore distinguish:

• unconstrained demand, or what customers would request if supply were unlimited;
• constrained shipments, or what the organization can actually deliver;
• incremental demand caused by an intervention;
• demand transferred from another product, period, or channel.

Failure to separate these concepts can cause a feedback loop: shortages reduce observed sales, the system interprets lower sales as lower demand, replenishment is reduced, and future shortages worsen.

4. S&OP and Integrated Business Planning

S&OP is a cross-functional process for reconciling demand, supply, inventory, capacity, and financial objectives. IBP extends this logic by linking operational plans more explicitly to strategic and financial planning. Neither process is merely a monthly meeting; each is a governance system with defined data, decision rights, escalation rules, and accountability.

A mature cycle commonly includes:

1. Data and portfolio review: assess history, changes, item status, and forecast performance.
2. Demand review: produce a baseline and document commercial assumptions.
3. Supply review: test capacity, materials, labor, logistics, and supplier constraints.
4. Reconciliation: evaluate gaps, scenarios, financial effects, and trade-offs.
5. Executive decision: approve priorities, resource changes, assumptions, and exceptions.
6. Execution and learning: communicate the plan, monitor deviations, and improve the process.

The central S&OP question is not “Which function is correct?” but “Which integrated plan best serves organizational objectives under uncertainty and constraint?” Effective governance makes disagreements explicit rather than hiding them in a single adjusted number.

A decision log should record the assumption, owner, evidence, expected effect, date of review, and consequence if the assumption fails. This improves organizational learning and reduces retrospective blame.

5. Demand Shaping and Constrained Planning

Forecasting traditionally treats demand as external. In practice, organizations can influence demand through price, promotions, assortment, lead-time promises, channel allocation, product configuration, substitution, and communication. Demand shaping is the deliberate use of such levers to align demand with profitable and feasible supply.

Examples include:

• shifting customers from a constrained product to an available substitute;
• offering a discount for off-peak delivery;
• limiting promotions when inventory is insufficient;
• prioritizing high-margin or strategically important orders;
• using postponement to delay final configuration;
• offering longer lead times in exchange for lower price.

Demand shaping must be incorporated into the forecast as an intervention with an expected response, not treated as an unexplained adjustment. Historical promotion effects can be estimated, but causal interpretation is difficult because promotions may coincide with holidays, competitor actions, or inventory availability.

Constrained planning begins with a demand signal and asks what can be supplied given finite resources. It should make trade-offs visible. A constrained plan may maximize contribution, service to strategic customers, total units, or another objective. These objectives are not equivalent.

A simplified contribution objective is:

subject to capacity and demand constraints, where p_i is selling price, c_i variable cost, x_i supplied quantity, and s_i unmet demand. This is an illustrative linear-planning structure, not a complete model. Real applications may include minimum service levels, integer production batches, setup times, inventory balance equations, contractual obligations, and multiple periods.

6. Cost-to-Serve and the Economics of Forecast Error

The consequences of forecast error should be evaluated through cost-to-serve. Relevant costs may include:

• purchase and production cost;
• inventory carrying cost;
• markdown and obsolescence cost;
• transportation and expediting cost;
• overtime and subcontracting;
• stockout, penalty, and lost-margin cost;
• customer dissatisfaction and future demand loss;
• carbon and environmental costs.

The economically preferred decision is not always the one with the lowest forecast error. For example, a slightly higher forecast may be justified for a critical spare part where stockout consequences are severe, while a conservative forecast may be appropriate for a perishable product with high waste cost.

A practical approach is to classify forecast errors by consequence. A dashboard can show accuracy, bias, service, inventory, waste, expedites, and margin together. This discourages local optimization, such as improving forecast accuracy by suppressing legitimate uncertainty or reducing inventory while causing costly shortages.

Forecast value can also be assessed by comparing a decision made with the forecast to a benchmark policy. The benchmark might be a seasonal naive forecast, existing planner judgment, or a simple replenishment rule. Improvement should be measured in business outcomes as well as statistical metrics.

7. Data, Technology, and AI-Enabled Forecasting

Modern forecasting may use point-of-sale data, orders, inventory availability, prices, promotions, weather, web traffic, macroeconomic indicators, and external market signals. Machine-learning models can capture nonlinear relationships and interactions, but complexity does not guarantee better decisions.

Technology should support, not replace, a disciplined process. Before deploying an advanced model, an organization should establish:

• a clear decision purpose and forecast horizon;
• reliable master data and event calendars;
• leakage-free training and testing procedures;
• appropriate temporal validation;
• explainability suitable for the decision and risk level;
• monitoring for drift, bias, and data quality;
• human ownership and override protocols.

Time-series validation must respect chronology. Randomly mixing future observations into a training set can produce unrealistically optimistic performance. Forecasts should be evaluated using rolling-origin or other time-ordered tests.

AI systems also create governance concerns. A model trained on historically biased allocation decisions may reproduce those biases. A model may infer sensitive customer characteristics from apparently neutral variables. Automated recommendations can obscure accountability, particularly when stock allocation affects essential goods, healthcare products, or vulnerable communities.

Human judgment should be added selectively. Unstructured overrides often degrade accuracy, but informed intervention is valuable when it introduces verified information absent from historical data. Every override should record its reason and later outcome so that judgment can be evaluated rather than treated as unquestionable expertise.

8. Illustrative Case: Northstar Components

Background

Northstar Components, a fictional industrial distributor, supplies replacement motors to food-processing plants. It carries 2,400 stock-keeping units through three regional warehouses. The company has experienced rising inventory, emergency shipments, and customer complaints despite an apparently improving forecast-accuracy percentage.

Problem

The organization used a single monthly forecast generated by sales managers. Sales targets, customer requests, and unconstrained demand were combined. The forecast was allocated equally across warehouses, even though regional demand patterns differed. Procurement then placed orders using a fixed two-month coverage rule.

An investigation found three problems. First, several large customers had experienced stockouts, so observed shipments understated true demand. Second, the accuracy metric excluded intermittent items with zero demand, making reported performance look better. Third, procurement had no visibility of the uncertainty range or supplier lead-time variation.

Analysis

Northstar segmented items using annual contribution, demand variability, criticality, and lead time. For a high-criticality motor, average weekly demand was 120 units, weekly standard deviation was 25 units, and lead time was four weeks with little variation. Using a 97.5% cycle-service target, z was approximately 1.96.

Expected lead-time demand was 120 × 4 = 480 units. Demand standard deviation during lead time was 25√4 = 50 units. Safety stock was 1.96 × 50 = 98 units, and the approximate reorder point was 480 + 98 = 578 units.

The calculation was then tested against actual supplier performance, minimum order quantities, warehouse capacity, and the difference between cycle service and fill rate. For intermittent, low-value items, Northstar used a different policy rather than applying the same normal-demand equation.

Approach and Solution

Northstar introduced a two-layer forecast. A statistical baseline was generated at item-region-week level, while commercial teams submitted documented event assumptions. The demand review separated baseline demand, promotion or project uplift, and demand lost through prior stockouts.

The supply review translated the consensus forecast into material requirements, labor needs, warehouse capacity, and supplier commitments. Critical components received differentiated service targets. Procurement negotiated postponement and cancellation rights for uncertain items, while transportation planners reserved capacity only for high-confidence peaks.

The company also introduced a cost-to-serve dashboard containing MAE, bias, fill rate, inventory, obsolescence, expedites, and margin. Overrides required an explanation and were reviewed after the relevant period.

Outcome

In this illustrative case, the first quarter after implementation produced lower emergency freight and improved service for critical motors, although total inventory did not fall uniformly. Inventory increased temporarily for a small number of strategically important items and declined for several low-value items that had previously been overstocked. Management judged the program successful because the trade-off improved contribution and customer reliability rather than merely lowering aggregate inventory.

Learning Points

The case demonstrates five principles:

1. Forecast accuracy must be measured at the level of the decision.
2. Stockouts can hide demand and corrupt future forecasts.
3. Safety stock depends on uncertainty and service objectives, not only average demand.
4. S&OP is a decision process, not a number-approval meeting.
5. Business outcomes should complement statistical accuracy metrics.

9. Implementation Framework and Best Practices

A practical forecast-to-decision system can be implemented through the following sequence:

1. Define the decision: Specify whether the output supports replenishment, capacity, pricing, staffing, or strategic investment.
2. Define the forecast object: Establish item, location, time bucket, horizon, unit, and whether demand is constrained or unconstrained.
3. Create a baseline: Use an appropriate statistical method and document data exclusions.
4. Add known events: Include promotions, launches, closures, contracts, and disruptions with evidence and ownership.
5. Represent uncertainty: Provide intervals, scenarios, or distributions where decisions are sensitive to risk.
6. Apply constraints: Test materials, capacity, lead times, labor, cash, storage, and service commitments.
7. Evaluate alternatives: Compare inventory, capacity, procurement, logistics, and demand-shaping actions using cost-to-serve.
8. Approve and communicate: Record decisions, assumptions, exceptions, and responsibilities.
9. Monitor outcomes: Track accuracy, bias, service, inventory, cost, and assumption validity.
10. Learn and recalibrate: Retire ineffective rules, revise segmentation, and improve data and models.

Best practice is not to seek one universal forecasting method. It is to match methods, aggregation, uncertainty treatment, and governance to the economics and characteristics of each decision.

Challenges and Limitations

Forecasting remains difficult because supply chains are adaptive systems. Prices change in response to forecasts; suppliers alter behavior after receiving commitments; customers bring orders forward when shortages are expected; and competitors react to market actions. Historical relationships may therefore change when a forecast becomes influential.

Additional limitations include structural breaks, new-product cold starts, sparse data, changing product hierarchies, demand substitution, unreliable master data, and inconsistent definitions of demand. Forecast reconciliation can enforce mathematical consistency across levels but cannot guarantee that the hierarchy itself is meaningful.

There is also a risk of false precision. A forecast stated as 10,247 units may imply more knowledge than the data support. Ranges and scenario narratives are often more honest and more useful for strategic decisions. Finally, organizations must avoid using forecast systems to disguise managerial choices. If leadership chooses to prioritize a customer or accept a lower service level, that is a policy decision and should be identified as such.

Chapter Summary

Forecasts become valuable when they are translated into explicit decisions under uncertainty. A forecast is distinct from a target, plan, budget, or scenario. Forecast error should be measured using appropriate metrics, with attention to bias, scale, intermittency, and business consequences.

Inventory decisions combine expected lead-time demand with safety stock derived from uncertainty and service objectives. Procurement, capacity, workforce, logistics, revenue, and network decisions require additional consideration of constraints, flexibility, lead times, and cost-to-serve. S&OP and IBP provide governance for reconciling demand, supply, financial objectives, and strategic priorities.

Demand shaping and constrained planning recognize that organizations can influence demand and must allocate scarce resources. AI and advanced analytics can improve forecasts when supported by sound data, time-ordered validation, monitoring, explainability, and accountable governance. The central managerial principle is to evaluate forecasts by the quality and economic value of decisions they enable, not by statistical accuracy alone.

Key Terms

Forecast: An estimate of a future outcome based on available information.

Forecast error: The difference between actual and forecast values.

Bias: Persistent over- or under-forecasting.

Baseline forecast: Expected demand before discretionary adjustments.

Consensus forecast: A cross-functional forecast incorporating validated information.

Plan: A selected course of action intended to achieve objectives.

Scenario: A coherent representation of a possible future.

Safety stock: Inventory held to protect against uncertainty.

Reorder point: Inventory position at which replenishment is triggered.

Cycle service level: Probability of avoiding a stockout during a replenishment cycle.

Fill rate: Proportion of demand fulfilled immediately from available stock.

Lead-time demand: Demand expected during replenishment lead time.

S&OP: Cross-functional process for reconciling demand, supply, inventory, and resources.

IBP: Integrated business planning linking operational and strategic-financial planning.

Demand shaping: Deliberate action to influence the timing, quantity, or composition of demand.

Constrained planning: Planning that incorporates finite materials, capacity, labor, or logistics.

Cost-to-serve: Total cost of fulfilling a customer, product, channel, or order requirement.

Forecast reconciliation: Adjustment of forecasts so that levels in a hierarchy are coherent.

Demand censoring: Incomplete observation of demand because supply was unavailable or restricted.

Forecast governance: Roles, controls, assumptions, decision rights, and monitoring used to manage forecasting.

Review Questions

Short Answer Questions

1. How does a forecast differ from a plan and a target?
2. Why can MAPE be unsuitable for intermittent demand?
3. What is the difference between cycle service level and fill rate?
4. Why should stockout information be considered when estimating demand?
5. Identify two ways in which procurement can respond to forecast uncertainty.

Long Answer Questions

1. Explain how a forecast can be translated into inventory, capacity, workforce, and logistics decisions. Discuss the additional information required beyond the point forecast.
2. Describe the stages of an effective S&OP or IBP process and explain how governance improves decision quality.
3. Compare statistical forecast accuracy with forecast value. Why might an organization accept a less accurate forecast?
4. Discuss demand shaping and constrained planning, including the risks of confusing unconstrained demand with shipments.
5. Evaluate the opportunities and governance risks associated with AI-enabled supply chain forecasting.

Analytical / Application Questions

1. Weekly demand averages 250 units with a standard deviation of 40 units. Lead time is fixed at four weeks and the target cycle-service level is 95% with z = 1.645. Calculate expected lead-time demand, safety stock, and reorder point. State the assumptions behind your calculation.
2. A retailer reports improved forecast accuracy but worsening stockouts and emergency freight. Diagnose at least four possible causes and propose a measurement and governance redesign.
3. A factory has forecast demand above available capacity for the next quarter. Develop a decision framework comparing overtime, subcontracting, postponement, customer prioritization, price changes, and lost sales. Identify the data needed to evaluate cost-to-serve trade-offs.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes a forecast from a plan?

A. A forecast is a desired outcome, while a plan is an unbiased estimate.

B. A forecast estimates a future outcome, while a plan specifies intended actions.

C. A forecast is always statistical, while a plan is always financial.

D. A forecast applies only to demand, while a plan applies only to supply.

Correct Answer: B. A forecast estimates a future outcome, while a plan specifies intended actions.

Question 2

If actual demand is 1,000 units and the forecast is 920 units, using e_t = A_t − F_t, what is the forecast error?

A. −80 units

B. 0 units

C. 80 units

D. 1,920 units

Correct Answer: C. 80 units

Question 3

Which metric is especially problematic when actual demand is zero?

A. MAE

B. RMSE

C. Mean error

D. MAPE

Correct Answer: D. MAPE

Question 4

Average weekly demand is 200 units, lead time is 3 weeks, and weekly demand standard deviation is 30 units. Under fixed lead time and independent demand, what is the approximate standard deviation of lead-time demand?

A. 30 units

B. 52 units

C. 90 units

D. 600 units

Correct Answer: B. 52 units

Question 5

Which statement about safety stock is correct?

A. It equals expected lead-time demand in every replenishment system.

B. It protects against uncertainty and depends on the service objective.

C. It eliminates the need to forecast demand.

D. It is determined only by selling price.

Correct Answer: B. It protects against uncertainty and depends on the service objective.

Question 6

What is a major purpose of S&OP?

A. To replace all statistical forecasts with sales targets

B. To reconcile demand, supply, resources, and financial objectives

C. To maximize inventory regardless of service requirements

D. To prevent executives from changing operational plans

Correct Answer: B. To reconcile demand, supply, resources, and financial objectives

Question 7

Which example represents demand shaping?

A. Recording a stockout as zero demand without adjustment

B. Increasing safety stock because lead time is uncertain

C. Offering customers a discount for delivery in an off-peak period

D. Measuring forecast error using MAE

Correct Answer: C. Offering customers a discount for delivery in an off-peak period

Question 8

Why can observed shipments understate true demand?

A. Shipments always include safety stock

B. Stockouts can prevent customer requests from becoming sales

C. Forecasts are always higher than orders

D. Lead time is unrelated to demand

Correct Answer: B. Stockouts can prevent customer requests from becoming sales

Question 9

Which validation approach is most appropriate for a time-series forecasting model?

A. Randomly mixing all observations before splitting the data

B. Using only the first observation for testing

C. Using time-ordered or rolling-origin validation

D. Testing only on the training data

Correct Answer: C. Using time-ordered or rolling-origin validation

Question 10

A company has limited capacity and must choose which orders to fulfill. Which factor is most directly relevant to a cost-to-serve decision?

A. Only the number of forecast observations

B. Contribution, shortage consequences, service commitments, and fulfillment costs

C. Only the product name

D. The color used in the forecast dashboard

Correct Answer: B. Contribution, shortage consequences, service commitments, and fulfillment costs