2. Demand Analysis, Forecasting and Inventory Classification

Introduction and Context

Inventory decisions begin with an understanding of demand. An organisation cannot decide how much to purchase, produce, store or replenish unless it has a reasoned view of what customers, internal users or downstream operations are likely to require. Demand analysis therefore precedes most inventory policies. It provides the evidence needed to balance two competing consequences: insufficient stock, which may cause lost sales and operational disruption, and excessive stock, which ties up capital and increases risks of obsolescence, damage and storage cost.

Demand is not always identical to sales. Sales may be constrained by stockouts, capacity limitations, price restrictions or distribution failures. Demand analysis attempts to identify the quantity that customers would require under prevailing conditions, while sales records show only what was actually fulfilled. This distinction is particularly important in logistics, where observed data may conceal unmet demand.

Forecasting converts demand information into an estimate of future requirements. Forecasts are not predictions of certainty; they are structured estimates subject to error. Effective inventory management does not seek a perfect forecast. It seeks a forecast that is sufficiently accurate, transparent and timely to support appropriate service, replenishment and risk decisions. Classification techniques then add another dimension by recognising that not every stock item deserves the same level of control. High-value, critical or unpredictable items require different policies from low-value, routine and stable items.

This chapter develops an integrated approach to demand analysis, forecasting and inventory classification. It examines demand patterns, forecasting methods, error measurement, classification frameworks and the managerial use of their combined results.

Learning Objectives

After studying this chapter, the reader should be able to:

1. distinguish different forms and patterns of demand relevant to inventory decisions;

2. explain the role, assumptions and limitations of qualitative, time-series and causal forecasting methods;

3. calculate and interpret common forecast-error measures;

4. select forecasting approaches according to data availability, demand characteristics and decision purpose;

5. apply ABC, XYZ, VED and FSN classification methods;

6. design a multicriteria classification scheme for differentiated inventory control;

7. evaluate the organisational, technological and ethical issues associated with demand forecasting; and

8. recommend inventory policies using forecast information and classification results together.

Demand as an Inventory Signal

Demand may be classified as dependent or independent. Independent demand arises from external customers or markets and is usually forecast from historical demand, market intelligence and commercial assumptions. Finished goods, spare parts sold to customers and retail products commonly have independent demand. Dependent demand is derived from the demand for another item. Components, packaging materials and subassemblies are examples. If a production plan requires 500 finished units and each unit uses two bearings, the gross dependent demand for bearings is 1,000 units before considering existing stock, scheduled receipts or yield losses.

This distinction affects the forecasting method. Independent demand is often forecast statistically or through market judgement. Dependent demand is more appropriately calculated from a master production schedule, bill of materials and inventory records, usually through material requirements planning. Forecasting a component independently when its usage is determined by a production schedule can create unnecessary error.

Demand can also be permanent, seasonal, trend-driven, cyclical, erratic or lumpy. Permanent demand has no strong systematic movement, although random variation remains. Seasonal demand follows a recurring pattern within a year, month, week or day. Trend demand changes persistently upward or downward. Cyclical demand is associated with longer economic or industry cycles. Erratic demand varies irregularly, while lumpy demand contains many zero-demand periods interrupted by relatively large transactions. Spare parts and project materials frequently display lumpy demand.

A further distinction is between base demand and promotional or exceptional demand. A temporary price discount, a competitor’s stockout, a one-time project or an emergency purchase may produce observations that should not be treated as normal recurring demand. Forecasting systems must identify such events rather than blindly extrapolate them.

Demand data should be examined at the level at which the decision is made. Daily demand may be volatile while weekly demand is stable; an aggregate product family may show a smooth pattern while individual stock-keeping units remain unpredictable. The relevant aggregation depends on supplier lead time, replenishment frequency, customer service requirements and the cost of being wrong.

From Data to Forecast: A Disciplined Process

A sound forecasting process begins with a clear decision question. The required forecast horizon, time interval and level of detail should be specified before selecting a method. A procurement decision may require a three-month monthly forecast, whereas warehouse labour planning may require a daily forecast for the next two weeks.

The process generally includes the following connected activities:

• define the item, location, time bucket and forecast horizon;
• collect sales, shipment, order, stockout, price, promotion and lead-time data;
• clean and reconcile the data, including missing values and abnormal observations;
• diagnose demand patterns and possible causal drivers;
• select and, where useful, compare forecasting methods;
• generate the forecast and an associated uncertainty estimate;
• monitor errors and bias after actual demand becomes available; and
• revise parameters, assumptions or the method when performance deteriorates.

Forecast governance is as important as mathematical technique. A forecast should have an owner, a documented source, a revision schedule and a process for incorporating commercial or operational judgement. Sales, marketing, procurement, production and finance may each hold partial knowledge. A cross-functional process can improve the forecast, but uncontrolled overrides may introduce optimism, political influence or double counting.

Qualitative Forecasting Methods

Qualitative methods are based primarily on informed judgement rather than a long, reliable numerical history. They are valuable for new products, structural market changes, unusual projects and situations in which historical data do not represent future conditions.

The executive opinion method combines views from senior managers. It is rapid but can be dominated by hierarchy or persuasive individuals. The sales-force composite gathers estimates from sales personnel who are close to customers. It may capture local intelligence but can be distorted by incentives, negotiation behaviour or limited knowledge of broader demand.

The Delphi method obtains independent estimates from a panel of experts through several rounds. Responses are normally summarised anonymously, allowing the panel to reconsider its views without direct pressure from senior participants. Delphi is useful where expert knowledge is dispersed, although it can be time-consuming and remains dependent on the quality of the panel.

Market research, customer surveys, test marketing and structured customer interviews can provide evidence about intentions and preferences. Stated intention, however, is not the same as realised purchase. Qualitative estimates should therefore be tested against comparable historical evidence, scenario analysis or pilot results whenever possible.

Judgement is not a substitute for analysis. A useful practice is to establish a statistical baseline and then require decision-makers to explain material adjustments in terms of identifiable events, such as a confirmed contract, a planned promotion or a known capacity change.

Time-Series Forecasting

Time-series methods use the historical sequence of demand observations. They are most useful when past patterns contain information about future demand and when the underlying environment is reasonably stable.

Naïve and Seasonal Naïve Forecasts

The simplest forecast assumes that the next period will equal the most recent actual demand:

F(t+1) = A(t)

where F(t+1) is the forecast for the next period and A(t) is actual demand in the current period. Despite its simplicity, the naïve method is an important benchmark. A complex model should not be adopted unless it improves meaningfully on this baseline.

A seasonal naïve forecast uses demand from the corresponding previous season. For monthly data with annual seasonality, the forecast for January may equal demand in the previous January. This approach is often effective when the seasonal pattern is stable but the organisation lacks sufficient data for a more elaborate model.

Moving Averages

A simple moving average forecasts the next period by averaging the most recent observations:

F(t+1) = [A(t) + A(t-1) + ... + A(t-n+1)] / n

A larger value of n smooths random variation but responds slowly to a genuine change. A smaller value responds more quickly but may produce a less stable forecast. A weighted moving average assigns greater weights to observations considered more relevant, with weights usually summing to one.

Exponential Smoothing

Single exponential smoothing updates the previous forecast using the latest error:

F(t+1) = αA(t) + (1 − α)F(t)

where α lies between zero and one. A high α gives more weight to recent demand and produces a responsive forecast. A low α produces greater smoothing. The parameter should be selected using validation data or a defensible error-minimisation procedure, not merely convenience.

For data with trend, Holt’s method estimates both level and trend. For data with trend and seasonality, Holt–Winters methods use level, trend and seasonal components. Additive seasonality is appropriate when seasonal fluctuations are roughly constant in absolute size; multiplicative seasonality is more suitable when seasonal variation grows with the level of demand.

Decomposition and Advanced Models

Time-series decomposition separates demand into components such as level, trend, seasonality and irregular variation. Decomposition assists interpretation even when it is not used as the final forecasting model. Autoregressive integrated moving average models and other statistical models may capture autocorrelation and complex patterns, but their value depends on data quality, parameter stability and the organisation’s ability to maintain them.

Machine-learning models can use lagged demand, calendar variables, promotions, weather, web traffic and other predictors. They may improve accuracy in high-volume environments, but they also increase requirements for data governance, model monitoring and explainability. A sophisticated model that is poorly maintained can be less useful than a transparent, well-calibrated baseline.

Causal Forecasting and Scenario Thinking

Causal methods explain demand through factors believed to influence it. Regression analysis may relate demand to price, income, advertising expenditure, economic indicators, customer population, service levels or weather. A simple linear form can be written as:

D = β₀ + β₁X₁ + β₂X₂ + ε

where D is demand, X variables are explanatory factors, β coefficients represent estimated relationships and ε captures unexplained variation.

Causal analysis is valuable when management expects a driver to change. For example, a price reduction may increase demand, while a planned shutdown may reduce it. Correlation alone does not establish causation. Analysts must consider omitted variables, reverse causality, multicollinearity and changes in customer behaviour. A relationship observed historically may not remain valid after a policy or market structure changes.

Scenario analysis complements statistical forecasting. A baseline scenario may represent the most likely conditions, while high- and low-demand scenarios represent plausible alternatives. The purpose is not to assign false precision but to support decisions about capacity, safety stock, supplier commitments and contingency actions. For strategic inventory, a range of outcomes may be more useful than a single point estimate.

Forecast Error: Measurement and Meaning

A forecast is evaluated by comparing forecast demand with actual demand after the period ends. Define error as:

e(t) = A(t) − F(t)

The sign convention must be stated because some organisations define error in the opposite direction.

Mean absolute error (MAE) is the average of absolute errors:

MAE = Σ|e(t)| / n

It is expressed in the units of demand and is easy to interpret. Mean squared error (MSE) averages squared errors and penalises large misses more heavily. Root mean squared error (RMSE) is the square root of MSE and returns the measure to the original unit.

Mean absolute percentage error (MAPE) is often written as:

MAPE = [Σ(|e(t)| / A(t)) / n] × 100

It is intuitive but problematic when actual demand is zero or very small. It can also treat a percentage error on a low-volume item as more significant than a large-unit error on a high-volume item. Alternatives include weighted absolute percentage error, symmetric MAPE and scale-dependent measures such as MAE.

Forecast bias indicates systematic over- or under-forecasting. A simple cumulative error is:

CFE = Σe(t)

A persistent positive or negative value suggests that the process is not merely experiencing random error. Bias may arise from sales incentives, deliberate padding, unrecognised seasonality, obsolete parameters or an assumption that demand will recover when evidence does not support it.

Error metrics must be interpreted in context. A small average error may conceal occasional severe failures that cause stockouts. Conversely, a forecast with somewhat larger error may be acceptable for a low-value item with short replenishment lead time. Forecast evaluation should therefore connect accuracy with business consequences, including service level, inventory investment, obsolescence and emergency transport.

Inventory Classification as Differentiated Control

Classification groups items according to characteristics relevant to control. It is not a description of the item’s identity; it is a decision device. The same item may belong to different categories under different criteria or at different locations.

ABC Analysis: Economic Significance

ABC analysis ranks items by annual consumption value, usually calculated as:

Annual consumption value = annual quantity used × unit cost

Items are then arranged from highest to lowest value and divided into A, B and C groups. A items account for a relatively high share of total value and a relatively small share of item count; C items account for a low share of value and a large share of item count. Exact percentage boundaries are managerial conventions rather than universal laws.

A items usually justify accurate records, frequent review, close supplier coordination, cycle counting and senior management attention. C items may be controlled through simpler reorder rules, larger order quantities or less frequent review, provided that their operational criticality is not high. ABC is a value-based method; it does not indicate whether demand is predictable or whether an item is essential.

XYZ Analysis: Demand Predictability

XYZ classification focuses on variability or forecastability. X items have relatively stable demand, Y items show moderate variation or trend, and Z items are highly irregular or difficult to forecast. Organisations may use the coefficient of variation, forecast error or another documented measure. The coefficient of variation is:

CV = standard deviation of demand / mean demand

Thresholds should reflect the organisation’s data and operating environment. XYZ analysis helps determine safety-stock logic, review frequency and the need for scenario planning. A Z item may require a risk-based order policy even if its average demand is low.

VED Analysis: Criticality of Consequences

VED classifies items as Vital, Essential or Desirable. A vital item’s absence may stop operations, threaten safety or create serious consequences. An essential item affects performance materially, although short-term alternatives may exist. A desirable item is useful but its absence causes limited disruption.

VED is especially relevant for maintenance, repair and operating supplies, hospitals, utilities and safety-related inventory. It requires operational and technical judgement. A low-cost fuse can be more critical than an expensive but easily substituted component.

FSN Analysis: Movement and Obsolescence

FSN groups items as Fast-moving, Slow-moving or Non-moving, based on issue frequency, quantity issued or time since last movement. The method supports warehouse layout, picking efficiency, ageing review and disposal decisions. Non-moving stock should not automatically be discarded; it may be required for emergency maintenance, contractual obligations or discontinued equipment. Its status should be reviewed with engineering, finance and operations.

Multicriteria Classification and Policy Design

Single-criterion classifications can conflict. An item may be low in annual value but vital to safety; it may be high value but highly predictable; or it may be fast-moving yet easy to replace. A multicriteria matrix combines ABC, XYZ, VED and sometimes FSN into a richer control profile, such as AX-V, CZ-E or BY-D.

The purpose is not to produce labels for their own sake. Each profile should lead to a policy. For example:

• AX-V items require close review, high record accuracy, validated forecasts, supplier-risk monitoring and strong contingency arrangements.
• CZ-D items may be managed with simple replenishment, standard storage and periodic review, unless other business constraints intervene.
• AZ-E items combine high value with uncertain demand and may require postponement, supplier agreements, scenario-based stock and management approval for purchases.
• CX-V items have low value but high criticality; maintaining availability may be rational despite their low financial contribution.

A classification policy should specify review frequency, forecasting method, service objective, safety-stock approach, approval authority, counting frequency and disposition rules. Classification should be refreshed periodically because value, demand, criticality and movement change over time.

Practical Application: Linking Forecasts to Inventory Decisions

Forecasting and classification should operate as a connected system. A forecast provides expected demand, while classification determines how much confidence, attention and protection the item deserves. High forecast accuracy does not eliminate the need for safety stock when lead time is long or the consequence of stockout is severe. Conversely, a highly critical item may need protection even when its forecast is poor.

For a stable X item, historical variation may support statistical safety-stock calculations. For a Z item, the organisation may rely more on criticality analysis, supplier lead-time risk, installed-base information and scenario planning. For an A item, forecast improvements can have a substantial financial effect because errors apply to a large consumption value. For a C item, the administrative cost of sophisticated forecasting may exceed the benefit.

Forecasts should also be reconciled across levels. Item forecasts should be consistent with product-family, market and capacity plans. When statistical forecasts and management judgement differ significantly, the difference should be documented and reviewed rather than hidden in an unexplained final number.

Illustrative Case: Spare-Parts Control at Meridian Process Equipment

Background

Meridian Process Equipment, a fictional manufacturer of industrial pumps, operates a central service warehouse supporting customers across three regions. Its inventory includes fast-moving seals, bearings and standard fittings, as well as specialised motor controllers and emergency replacement assemblies. The company had rising inventory value but continued to experience urgent shipments and customer downtime.

Situation and Analysis

An initial review showed that the planning team used a single monthly forecast method for all spare parts. Items with intermittent demand were forecast using averages that frequently produced zero or excessive requirements. The team also used ABC analysis alone. As a result, a low-cost but vital controller component received little attention, while some high-value but easily substituted items were overprotected.

The company cleaned two years of issue data and separately recorded known stockout periods. It classified items by annual consumption value, demand variability, operational criticality and movement. The analysis identified four important groups:

1. high-value, stable-demand bearings requiring accurate replenishment and frequent review;

2. high-value, irregular motor controllers requiring scenario-based planning and supplier coordination;

3. low-value, vital seals that needed high availability despite low financial value; and

4. non-moving assemblies that required engineering review before disposal because some were linked to installed equipment still under service contracts.

Approach and Strategy

For stable items, Meridian used weighted moving averages and compared them with exponential smoothing. For seasonal service demand, it tested seasonal naïve forecasts. For intermittent items, it stopped treating every zero as evidence of permanently low demand and used installed-base data, failure rates, supplier lead times and criticality assessments. Forecast performance was monitored with MAE and bias rather than MAPE alone, because several items had zero-demand periods.

The resulting control policies differentiated review frequency and service protection. Vital items received minimum availability targets and approved emergency alternatives. A items were cycle-counted more frequently, while selected C items were managed through simpler periodic review. Non-moving stock was reviewed quarterly by operations, engineering and finance.

Outcome

The main improvement was not a single forecasting formula. It was the alignment of data quality, item classification and replenishment policy. Planners could explain why a forecast was used, which items required judgement and which risks justified holding stock. Emergency shipments declined during the subsequent evaluation period, while management gained better visibility of obsolete and strategically necessary inventory. The case illustrates that forecasting accuracy and classification effectiveness are complementary rather than competing objectives.

Learning Points

The case yields four lessons. First, stockout observations must be interpreted carefully because recorded demand may be censored by unavailable stock. Second, financial value alone cannot represent operational importance. Third, intermittent demand requires methods and information different from those used for stable fast-moving products. Finally, classification is useful only when it changes a real control decision.

Technology, Analytics and Governance

Modern warehouse and enterprise systems provide transaction histories, inventory positions, purchase orders, customer orders and supplier performance data. Forecasting applications can automate model selection, detect outliers, calculate error measures and generate alerts for bias. Demand-sensing tools may incorporate near-real-time signals such as orders, web activity, weather or transport information.

Artificial intelligence and machine learning can identify nonlinear relationships and interactions across many variables. Their use should be proportionate to the decision. Organisations need version control, data lineage, access controls, monitoring for drift and a clear process for human review. An algorithm trained on stock-constrained sales may learn historical under-fulfilment rather than true demand. Similarly, automated classifications may reproduce biased or incomplete criticality data.

Governance requires ownership of master data, item descriptions, units of measure, costs, lead times and classification thresholds. A change to a unit cost can alter ABC status; a new production line can change VED status; and a supplier change can alter XYZ risk. Audit trails are important where inventory decisions affect safety, regulated products or significant financial reporting.

Challenges and Common Mistakes

Several recurring mistakes weaken demand and classification practice. The first is confusing sales with unconstrained demand. The second is using one forecasting method for every item. The third is selecting a model by mathematical sophistication rather than decision value. The fourth is measuring only average accuracy while ignoring bias, stockout consequences and service performance.

Other problems include excessive forecast overrides, unstable classification thresholds, failure to distinguish genuine outliers from data errors, ignoring lead-time variability and treating non-moving inventory as automatically useless. Aggregating items too early can conceal item-level risk, while analysing every item with equal sophistication can waste scarce analytical resources.

Forecasts also create behavioural risks. Managers may inflate forecasts to secure capacity or reduce them to meet inventory targets. Suppliers may receive inconsistent signals when internal forecasts are not reconciled. Ethical practice requires transparent assumptions, evidence-based overrides and honest communication of uncertainty.

Emerging Directions and Best Practices

Demand planning is moving toward probabilistic forecasting, in which a range or distribution is estimated instead of only a single expected value. This supports decisions involving service levels and safety stock more directly. Collaborative forecasting increasingly combines supplier, customer and internal data, while digital twins and advanced simulation allow organisations to test disruption scenarios before committing inventory.

Useful best practices include:

• begin with a clear decision and forecast horizon;
• establish a transparent statistical baseline;
• segment items before investing in advanced methods;
• measure error and bias at the level of the relevant decision;
• treat stockouts, promotions and abnormal events explicitly;
• link ABC, XYZ, VED and FSN results to written control policies;
• review classifications when business conditions change; and
• combine automation with accountable professional judgement.

Chapter Summary

Demand analysis provides the foundation for inventory decisions by distinguishing actual sales from potential demand and by identifying permanent, seasonal, trend, erratic and lumpy patterns. Independent demand is generally forecast, whereas dependent demand is derived from production or usage requirements. Qualitative methods are useful when data are limited or structural change is significant; time-series methods exploit historical patterns; and causal methods relate demand to explanatory variables. Forecasts must be evaluated through error, accuracy and bias measures, with attention to business consequences rather than a single metric.

ABC classification reflects annual consumption value, XYZ reflects demand predictability, VED reflects operational criticality and FSN reflects movement. Because these dimensions answer different questions, multicriteria classification is often more useful than any one method alone. The ultimate purpose of classification is differentiated control: forecasting effort, review frequency, service protection, counting, purchasing and disposition should correspond to the item’s financial, operational and demand risks. Digital analytics can strengthen this process, but data quality, governance, explainability and human accountability remain essential.

Key Terms

Demand: The quantity of a product or service that users are willing and able to require under specified conditions.

Independent demand: Demand arising from external customers or markets and not directly determined by another inventory item.

Dependent demand: Demand derived from the requirement for another product, component or production plan.

Forecast: A structured estimate of future demand based on information and assumptions.

Forecast horizon: The future time period covered by a forecast.

Time-series method: A forecasting approach that uses the historical sequence of observations.

Causal forecasting: Forecasting that relates demand to explanatory variables such as price, promotion or economic conditions.

Seasonality: A recurring demand pattern associated with a particular period or calendar cycle.

Exponential smoothing: A method that updates a forecast by weighting recent actual demand and the previous forecast.

Forecast error: The difference between actual demand and forecast demand for a period.

MAE: Mean absolute error, the average absolute difference between actual and forecast demand.

Forecast bias: Persistent tendency to over-forecast or under-forecast.

ABC analysis: Classification based on annual consumption value.

XYZ analysis: Classification based on demand variability or predictability.

VED analysis: Classification into vital, essential and desirable items according to consequence of non-availability.

FSN analysis: Classification into fast-moving, slow-moving and non-moving items.

Coefficient of variation: Standard deviation divided by mean demand, used as a relative variability measure.

Multicriteria classification: Classification that combines two or more dimensions, such as value, variability and criticality.

Demand sensing: Use of recent operational or market signals to update demand estimates.

Safety stock: Additional inventory held to protect against demand or supply uncertainty.

Review Questions

Short Answer Questions

1. How does dependent demand differ from independent demand?

2. What is the purpose of a naïve forecast as a benchmark?

3. Define forecast bias and explain why it matters.

4. What does VED classification measure?

5. Why can MAPE be unsuitable for intermittent demand?

Long Answer Questions

1. Explain the main demand patterns relevant to inventory management and discuss their implications for forecasting.

2. Compare qualitative, time-series and causal forecasting methods, including the circumstances in which each is appropriate.

3. Describe exponential smoothing, moving averages and seasonal forecasting, and explain the trade-off between responsiveness and stability.

4. Explain ABC, XYZ, VED and FSN analyses and show why no single classification is sufficient for all inventory decisions.

5. Discuss how forecast-error measurement, governance and technology can improve inventory decision-making.

Analytical / Application Questions

1. A spare part has demand of 10, 0, 12, 0 and 18 units over five successive months. Explain why a simple average may be misleading and recommend information that should supplement the forecast.

2. An item represents only 2 percent of annual inventory value but its absence stops a safety-critical production line. Which classification dimension is likely to override its ABC status, and what control policy would you recommend?

3. A planner reports excellent MAPE for a group containing many zero-demand periods, but stockouts remain frequent. Analyse why the metric may be misleading and propose a better evaluation approach.

22. Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes independent demand from dependent demand?

A. Independent demand is always stable, while dependent demand is always variable.

B. Independent demand arises from external requirements, while dependent demand is derived from another item or production plan.

C. Independent demand applies only to raw materials, while dependent demand applies only to finished goods.

D. Independent demand is never forecast, while dependent demand is always forecast statistically.

Correct Answer: B. Independent demand arises from external requirements, while dependent demand is derived from another item or production plan.

Question 2

A forecasting team wants a method that gives substantially greater weight to the most recent observation. Which adjustment is most appropriate in single exponential smoothing?

A. Increase the smoothing constant α.

B. Decrease the smoothing constant α.

C. Increase the number of historical periods averaged.

D. Replace actual demand with annual consumption value.

Correct Answer: A. Increase the smoothing constant α.

Question 3

Why should a seasonal naïve forecast be considered for a product with stable annual seasonality?

A. It eliminates all forecast error.

B. It uses demand from the corresponding prior season as a practical benchmark.

C. It is designed only for products with no historical data.

D. It automatically explains the causal effect of price changes.

Correct Answer: B. It uses demand from the corresponding prior season as a practical benchmark.

Question 4

Which measure is most directly intended to identify systematic over-forecasting or under-forecasting?

A. Mean absolute error

B. Forecast bias or cumulative forecast error

C. Annual consumption value

D. Coefficient of variation alone

Correct Answer: B. Forecast bias or cumulative forecast error

Question 5

Why can MAPE be problematic for intermittent-demand items?

A. It cannot be expressed as a percentage.

B. It ignores all forecast errors larger than one unit.

C. Zero or very small actual values can make percentage errors undefined or disproportionately large.

D. It measures only supplier lead-time variability.

Correct Answer: C. Zero or very small actual values can make percentage errors undefined or disproportionately large.

Question 6

An item has low annual consumption value but its absence would create a serious safety risk. Which classification is most important in this situation?

A. VED

B. FSN

C. ABC only

D. Seasonal naïve classification

Correct Answer: A. VED

Question 7

What is the primary basis of ABC analysis?

A. Demand criticality

B. Demand predictability

C. Annual consumption value

D. Time since last movement only

Correct Answer: C. Annual consumption value

Question 8

A product has high value and highly irregular demand. Which combined profile is most appropriate?

A. AX

B. AY

C. AZ

D. CZ

Correct Answer: C. AZ

Question 9

Which action best reflects sound forecast governance?

A. Allow every department to change the forecast without recording reasons.

B. Use one forecasting method for all stock-keeping units.

C. Document assumptions, assign ownership and monitor forecast error and bias.

D. Judge forecast quality only by whether total inventory value declines.

Correct Answer: C. Document assumptions, assign ownership and monitor forecast error and bias.

Question 10

Why may recorded sales underestimate true demand?

A. Sales records always include promotional demand twice.

B. Stockouts or capacity constraints may prevent some desired demand from being fulfilled.

C. Sales data automatically measure customer intentions.

D. Warehouse movement data include no information about availability.

Correct Answer: B. Stockouts or capacity constraints may prevent some desired demand from being fulfilled.