Demand Planning, Forecasting, and Inventory Software

Introduction

Demand planning and inventory management are among the most software-intensive activities in supply chain management. Modern supply chain management (SCM) platforms integrate historical transactions, point-of-sale data, customer orders, promotions, market signals, inventory records, supplier constraints, and logistics information into a coordinated planning environment. Their purpose is not merely to produce a forecast. They support decisions about what demand is likely to occur, how much inventory should be held, when replenishment should occur, and how alternative business assumptions may affect service, cost, and working capital.

Software does not eliminate uncertainty. Instead, it makes uncertainty visible, quantifiable, and manageable. A planner must therefore understand both the mathematics embedded in the system and the managerial judgments applied around it. A statistically accurate forecast can still produce poor decisions if lead times are wrong, inventory records are inaccurate, product substitutions are ignored, or a planner overrides the system without evidence. Conversely, a sophisticated algorithm cannot compensate for weak governance or unreliable master data.

This chapter examines the principal software capabilities used for demand planning, forecasting, inventory segmentation, safety-stock calculation, replenishment, and multi-echelon inventory management. It also explains forecast error, bias, service levels, scenarios, and planner overrides from both quantitative and managerial perspectives.

Learning Objectives

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

1. Explain the role and architecture of demand-planning and inventory software.
2. Distinguish demand, sales, orders, shipments, and consumption data.
3. Compare major forecasting approaches and interpret forecast accuracy and bias measures.
4. Apply segmentation methods such as ABC, XYZ, and demand-pattern classification.
5. Calculate basic reorder points and safety stocks under stated assumptions.
6. Explain service-level trade-offs and the logic of multi-echelon inventory planning.
7. Evaluate scenarios, overrides, exceptions, and software-generated recommendations.
8. Identify data, governance, ethical, and implementation risks in SCM planning systems.

1. The Role of Software in Demand and Inventory Decisions

A typical planning platform connects several data and decision layers:

The central distinction is between a transaction system and a planning system. Enterprise resource planning (ERP) systems record orders, receipts, production, and inventory movements. Advanced planning systems (APS) use these records, often with additional external and commercial data, to evaluate future possibilities. The planning system may recommend an order, but execution normally remains subject to procurement, production, financial, and operational controls.

An integrated process commonly follows this sequence:

1. cleanse and reconcile data;
2. construct a baseline forecast;
3. incorporate known commercial events and causal information;
4. segment products, locations, and customers;
5. calculate inventory targets and replenishment proposals;
6. run scenarios and review exceptions;
7. approve and publish a plan;
8. monitor actual results and learn from forecast error.

This process is iterative rather than purely sequential. A promotion decision can change the forecast; the forecast can change capacity requirements; capacity limitations can then change the feasible supply plan.

2. Demand Concepts and Data Requirements

Demand is the quantity that customers would purchase under relevant market conditions. It is not always identical to sales, shipments, or orders. If an item is out of stock, observed sales may be lower than true demand. If a distributor builds inventory, shipments may temporarily exceed end-customer consumption. A planning system should therefore distinguish at least the following measures:

• Orders: customer requests placed during a period.
• Shipments: quantities dispatched by a supplying node.
• Sales: transactions recognized commercially or financially.
• Consumption: usage by the final customer or downstream operation.
• Lost sales: demand that was not fulfilled because of stockout or another constraint.
• Dependent demand: demand derived from a production plan, bill of material, or higher-level item.
• Independent demand: demand determined primarily by external customers or markets.

Data quality strongly affects every downstream calculation. Important fields include item and location identifiers, units of measure, calendars, product hierarchies, historical prices, promotions, stockout flags, lead times, minimum order quantities, order multiples, supplier reliability, and inventory status. A system may calculate precisely from inaccurate data; precision should not be confused with validity.

A demand history should be adjusted carefully for abnormal events. A one-time liquidation sale, a plant closure, or a stockout may not represent ordinary future demand. However, removing every unusual observation can be equally harmful because genuine structural changes may be incorrectly classified as noise. Adjustments should be documented, reversible, and governed.

3. Forecasting in Planning Software

3.1 Forecasting logic

A forecast is an estimate of future demand conditional on information available at the time it is produced. Forecasts are not promises and should not be judged solely by whether the actual result equals the prediction. A useful system records the forecast vintage: the forecast that was available before the period occurred. Comparing actual demand with a later, manually revised forecast can conceal the original error.

Common approaches include:

• Naïve forecast: the next period equals the most recent observed value. It is a useful benchmark.
• Moving average: the forecast is the arithmetic mean of recent observations.
• Exponential smoothing: recent observations receive more weight, with weights declining over time.
• Trend and seasonal models: systematic growth, decline, and recurring seasonal patterns are estimated.
• Causal models: demand is related to explanatory variables such as price, promotion, weather, or economic activity.
• Intermittent-demand methods: specialized methods address many zero-demand periods separated by sporadic demand.
• Machine-learning models: algorithms identify nonlinear relationships and interactions across large data sets.

For a simple moving average of the last n observations, the forecast for the next period is:

Here, F_t+1 is the forecast for period t+1, A_t-i is actual demand in a prior period, and n is the number of observations. The method assumes that recent observations are representative and that no explicit trend or seasonality requires adjustment. All observations must use the same unit and time basis.

Simple exponential smoothing is commonly written as:

where 0 < α ≤ 1 is the smoothing constant, A_t is actual demand in period t, and F_t is the forecast for period t. A larger α reacts more quickly to changes but may produce a less stable forecast. Basic exponential smoothing is appropriate when level is present but systematic trend and seasonality are absent; extensions are needed when those patterns exist.

Forecast selection should be based on out-of-sample performance, operational relevance, explainability, and stability—not merely on the most complex algorithm. A model that is marginally more accurate but impossible for planners to explain may be less useful in a regulated or highly collaborative environment.

3.2 Forecast error and bias

Forecast error for a period is commonly defined as:

Positive error means actual demand exceeded the forecast; negative error means the forecast exceeded actual demand. A planning organization must define its sign convention explicitly because some systems use forecast minus actual.

The mean absolute error (MAE) is:

MAE expresses the average absolute error in the same units as demand. The mean absolute percentage error (MAPE) is:

MAPE is unsuitable when actual demand is zero and can become unstable when actual demand is very small. A system should then use alternatives such as weighted absolute percentage error, mean absolute scaled error, or measures based on volume-weighted error.

The mean error (ME) is:

ME indicates systematic bias. A positive ME under the convention above indicates underforecasting; a negative ME indicates overforecasting. Accuracy and bias are different. Positive and negative errors can cancel in ME while remaining large in absolute terms. Therefore, a forecast should normally be monitored using both an error-magnitude measure and a bias measure.

A useful operational measure is the tracking signal:

where RSFE = ∑e_t is the running sum of forecast errors and MAD = (1/n)∑|e_t| is mean absolute deviation over the selected monitoring window. Tracking-signal thresholds are governance rules rather than universal laws; they should be calibrated to the organization’s data and decision cycle.

3.3 Forecast value and judgment

Forecast value analysis compares the performance of a baseline statistical forecast with the final forecast after human intervention. If overrides consistently reduce error, the intervention process may add value. If they increase error, the organization may have an optimism, anchoring, or political-bias problem. A fair evaluation must preserve the baseline forecast and record the reason, author, time, magnitude, and approval status of each override.

Judgment is valuable when planners possess information not available in the data, such as a confirmed customer closure or an unannounced regulatory change. Judgment is less defensible when it simply reflects a sales target or an unsupported belief that “the market will improve.” Software should make the distinction visible rather than hide it.

4. Segmentation and Inventory Policies

No single inventory policy is appropriate for every item. Segmentation assigns differentiated attention, service objectives, forecasting methods, and replenishment rules.

4.1 ABC analysis

ABC analysis classifies items by annual consumption value:

Items are ranked from highest to lowest value and divided into managerial categories. The category boundaries are policy choices, not natural laws. A-items usually receive close review and accurate master data; C-items may use simpler controls and larger review intervals. ABC analysis alone does not identify demand variability or strategic criticality.

4.2 Variability and criticality

XYZ analysis classifies demand predictability. X items have relatively stable demand, Y items may show trend or seasonality, and Z items are highly irregular. Another dimension is criticality: a low-value spare part may be operationally indispensable even when its annual consumption value is small. A robust segmentation design may therefore combine value, variability, lead time, margin, customer importance, and failure consequence.

Segmentation should lead to differentiated policies, for example:

Segmentation is dynamic. A product should be reclassified when its demand, cost, lead time, or business role changes.

5. Safety Stock, Reorder Points, and Service Levels

5.1 Reorder point

A reorder point is the inventory position at which replenishment should be initiated. Under a simple continuous-review model with constant demand and lead time:

where μ_L is expected demand during lead time and SS is safety stock. If average demand per period is d and lead time is L periods, then:

These relationships assume consistent units, a defined inventory-position calculation, and a replenishment process in which orders are triggered when inventory position reaches the reorder point. Inventory position generally includes on-hand inventory plus scheduled receipts minus backorders or allocations, according to the organization’s policy.

5.2 Safety stock with demand variability

If demand during lead time is approximately normally distributed and lead time is fixed, safety stock for a cycle-service target can be represented as:

Here, σ_L is the standard deviation of demand during lead time and z is the standard-normal quantile corresponding to the desired probability of avoiding a stockout during the replenishment cycle. For example, a 95% cycle-service target uses approximately z = 1.645, subject to the assumptions of the model.

If period demand is independent with standard deviation σ_d and lead time is L periods, then:

This formula assumes fixed lead time and independent, identically distributed demand. It should not be used without adjustment where demand is autocorrelated, lead time is variable, or demand and lead time are correlated.

When both demand and lead time vary, a commonly used approximation is:

where d is average demand per period, σ_d is the standard deviation of demand per period, L is average lead time, and σ_L is the standard deviation of lead time. This approximation assumes independence between demand and lead time and is most defensible when the distributions and operating conditions make the approximation reasonable. If demand and lead time are correlated, a covariance term is required:

The exact model used by commercial software may differ, particularly for periodic-review systems, non-normal demand, intermittent demand, service-level definitions, and finite replenishment rates. Users should inspect the parameter documentation rather than assume that every “safety stock” field has the same meaning.

5.3 Worked example

A retailer sells an item at an average rate of 100 units per week. Weekly demand has a standard deviation of 20 units. Supplier lead time is fixed at 3 weeks. The retailer wants a 95% cycle service level.

Expected lead-time demand is:

The standard deviation during lead time is:

Using z = 1.645, safety stock is:

Therefore, the reorder point is:

This result is not a guarantee that every cycle will avoid a stockout. It is a statistical target based on the stated assumptions. If demand is strongly seasonal, lead time varies, or the demand distribution is highly skewed, the software should use an appropriate alternative model.

5.4 Service-level terminology

A cycle service level is the probability of not experiencing a stockout during a replenishment cycle. A fill rate is the proportion of demand fulfilled immediately from available inventory. These measures are related but not identical. A cycle can experience one stockout and still achieve a high fill rate if the shortage is small; alternatively, a single large shortage can reduce fill rate substantially.

Higher service levels generally require more inventory, but the relationship is nonlinear. Moving from 90% to 95% may require a modest increase, whereas moving from 98% to 99.5% may require disproportionately more inventory. Managers should therefore select service targets by customer value, margin, criticality, and shortage consequence rather than applying one universal target.

6. Replenishment and Multi-Echelon Inventory

Replenishment software converts demand and inventory policies into purchase, production, transfer, or allocation recommendations. Common policy types include:

• Continuous review: monitor inventory position continuously and order when a reorder point is reached.
• Periodic review: review at fixed intervals and order enough to reach a target level.
• Min–max policy: replenish when inventory falls below a minimum toward a maximum level.
• Order-up-to policy: order to a target position after each review.
• Lot-for-lot or requirement-based planning: order according to calculated requirements, often for dependent demand.

Practical constraints include minimum order quantities, pack sizes, truck capacities, supplier calendars, production batch sizes, shelf life, and budget limits. A mathematically attractive recommendation may be infeasible if it violates these constraints.

6.1 Multi-echelon logic

A multi-echelon inventory system contains several connected stocking points, such as suppliers, factories, regional distribution centers, and stores. Optimizing each node independently can produce excessive total inventory because every node protects itself against the same uncertainty. Multi-echelon inventory optimization considers the network, lead-time relationships, demand aggregation, postponement, and service requirements.

Inventory may be positioned upstream when demand is uncertain but the product can be customized late. It may be positioned downstream when customer response time is critical or transportation lead time is long. The appropriate location depends on total landed cost, service requirements, manufacturing flexibility, and risk pooling.

A system-wide design must distinguish local service from end-customer service. A distribution center can achieve a high local service level while stores still experience shortages if allocation rules, transportation schedules, or store-level variability are ignored. Multi-echelon software therefore requires accurate network structure and careful interpretation of which service metric is being optimized.

7. Scenarios, Exceptions, and Planner Overrides

A scenario is a controlled alternative set of assumptions used to assess consequences before a decision is implemented. Typical scenarios include a promotion, supplier disruption, demand decline, capacity reduction, tariff change, or altered service target. A scenario should specify its baseline, changed assumptions, time horizon, affected products and locations, and evaluation metrics.

Useful scenario outputs include forecast volume, inventory investment, projected shortages, capacity utilization, working capital, revenue at risk, and expedited freight. Scenarios should not be treated as predictions with equal probability unless probabilities have been explicitly estimated. They are decision-support experiments.

Exception management directs attention to conditions requiring review, such as:

• forecast bias beyond an approved threshold;
• inventory below safety stock;
• projected stockout before replenishment arrival;
• unusual demand or order quantity;
• supplier lead-time deterioration;
• excess or obsolete inventory;
• a recommendation violating a minimum order or capacity constraint.

Effective exception management avoids both alert fatigue and silent failure. Thresholds should reflect materiality, risk, and the planner’s ability to act.

A planner override should be recorded with the original value, revised value, reason, evidence, author, date, and approval status. Overrides should be evaluated after the relevant period. Frequent overrides may indicate a deficient model, poor data, a recurring business process not represented in the system, or inappropriate organizational incentives.

8. Illustrative Case: Northstar Medical Supplies

Background

Northstar Medical Supplies distributes sterile procedure kits through one national distribution center and four regional facilities. The company historically used spreadsheets and fixed minimum-stock values. Stockouts of high-criticality kits caused emergency shipments, while slow-moving kits accumulated and occasionally expired.

Problem

Management implemented demand and inventory software, but the first system-generated plan produced surprising results. The baseline forecast underpredicted demand for one hospital network, safety stock was excessive for several stable items, and regional facilities appeared to duplicate protection against the same demand uncertainty.

Analysis

A cross-functional team first reconciled item identifiers, units of measure, inventory status, and lead-time records. It separated hospital consumption from distribution-center shipments and flagged stockout periods. Items were segmented using annual consumption value, demand variability, criticality, and shelf life. Forecast accuracy was measured using MAE, while bias was monitored using cumulative error. The team found that sales managers had repeatedly increased forecasts to reflect targets, creating systematic overforecasting for several categories.

The inventory model was then reviewed. The distribution center used fixed lead times even though supplier performance varied significantly. Regional safety stocks were calculated independently, causing duplicated protection. The team introduced differentiated service targets: critical kits received high protection, while noncritical consumables used lower targets and more economical replenishment.

Approach and solution

Northstar established a baseline statistical forecast and required documented evidence for overrides. Promotion and contract changes were entered as explicit events rather than embedded informally in manual adjustments. Supplier lead times were updated from receipt history, and the network model evaluated inventory at the national and regional levels together. Planners reviewed exceptions weekly, while master-data owners reviewed parameters monthly.

Outcome

In the illustrative scenario, Northstar reduced emergency shipments and excess inventory simultaneously over two planning cycles. The improvement did not come from a single algorithm. It resulted from better demand definitions, segmentation, service-level differentiation, lead-time governance, and disciplined override review. Critical-item availability improved even though total inventory declined.

Learning points

1. Inventory reduction and service improvement are not necessarily contradictory when duplicated buffers and poor segmentation are removed.
2. A forecast model cannot correct distorted demand history without appropriate data treatment.
3. Service levels must reflect business consequence, not only item value.
4. Multi-echelon analysis is essential when several facilities protect the same customer demand.
5. Implementation success depends on governance and planner behavior as much as on software functionality.

9. Implementation, Governance, and Ethical Considerations

Implementation should begin with process and data design rather than software configuration alone. Organizations should define ownership for item attributes, lead times, calendars, service policies, forecast hierarchies, and exception thresholds. A data dictionary should specify definitions such as “available inventory,” “forecast error,” and “service level.”

Model governance includes version control, validation, monitoring, change approval, and retirement rules. Forecast models should be tested on historical holdout periods or rolling-origin evaluations rather than judged only on the data used to fit them. Users should know when an algorithm has changed and whether a comparison remains valid.

AI and machine learning can improve forecasting when abundant, relevant, and timely data exist. They can also amplify data leakage, historical bias, and false confidence. A model trained on past allocation decisions may learn that certain customers receive less supply and incorrectly interpret this as low demand. Human oversight is required where decisions affect essential goods, vulnerable customers, or contractual fairness.

Important ethical and governance issues include:

• protecting commercially sensitive customer and supplier data;
• controlling access to forecasts, prices, and inventory positions;
• explaining automated recommendations to affected users;
• preventing sales incentives from corrupting demand signals;
• avoiding discriminatory allocation when shortages occur;
• retaining audit trails for overrides and approvals;
• complying with applicable privacy, record-retention, and sector regulations.

The appropriate principle is not “automate everything.” It is to automate repeatable calculations, expose assumptions, preserve accountability, and reserve informed human judgment for decisions involving ambiguity, exceptional risk, or competing stakeholder obligations.

10. Best Practices

A mature demand and inventory software capability generally follows these practices:

1. Establish a trusted, common data model before optimizing algorithms.
2. Separate demand history from sales targets and supply-constrained shipments.
3. Preserve baseline forecasts and measure the value of overrides.
4. Monitor both error magnitude and directional bias.
5. Use segmentation to differentiate methods, review frequency, and service targets.
6. Document assumptions behind safety-stock and service-level calculations.
7. Include order multiples, capacity, shelf life, and supplier calendars in replenishment logic.
8. Evaluate inventory across the network rather than optimizing nodes in isolation.
9. Use scenario analysis for uncertainty instead of embedding arbitrary adjustments.
10. Design exception alerts around actionable risk and materiality.
11. Review model performance by product, location, horizon, and demand pattern.
12. Treat software outputs as recommendations subject to governance, not unquestionable instructions.

Challenges and Limitations

Forecasting remains difficult when products have short life cycles, intermittent demand, structural market changes, or weak causal data. Historical models can fail during pandemics, geopolitical disruptions, regulatory changes, and major channel shifts. Forecast accuracy metrics can also create perverse incentives: a manager may reduce forecast error by producing a conservative forecast, yet cause preventable stockouts.

Safety-stock formulas based on normality and independence are useful approximations, not universal truths. Service-level definitions may be misunderstood, and system parameters can become stale. Multi-echelon optimization can be computationally and organizationally demanding because decisions at one node affect others. Finally, sophisticated software can create a false impression of objectivity when assumptions and data-quality problems remain hidden.

Chapter Summary

Demand planning and inventory software connects data, forecasting, inventory policy, replenishment, collaboration, and execution. Its effectiveness depends on clear definitions of demand, reliable master data, appropriate forecasting methods, and disciplined governance. Forecast error should be measured using both magnitude and bias, while metrics such as MAPE must be applied cautiously when demand is zero or small.

Segmentation enables differentiated treatment of products according to value, variability, criticality, and other business characteristics. Reorder points combine expected lead-time demand with safety stock, but safety-stock calculations are valid only under stated assumptions about demand, lead time, distributions, and service definitions. Replenishment must also respect operational constraints.

Multi-echelon inventory planning considers the network rather than treating each stocking point independently. Scenarios reveal the consequences of alternative assumptions, and planner overrides should be evidence-based, documented, and evaluated for forecast value. AI may strengthen planning where data and governance are adequate, but it does not remove the need for accountability, ethical allocation, and professional judgment.

Key Terms

Demand planning: The process of translating market and operational information into an anticipated demand plan.

Forecast: An estimate of future demand based on information available at a specified time.

Forecast error: The difference between actual demand and forecast demand under a defined sign convention.

Forecast bias: Persistent directional tendency to overforecast or underforecast.

MAE: Mean absolute error, the average absolute forecast error in demand units.

MAPE: Mean absolute percentage error, a percentage error measure that is problematic when actual demand is zero or very small.

Safety stock: Inventory held to protect against demand, lead-time, or supply uncertainty.

Reorder point: Inventory-position threshold that triggers replenishment.

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

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

Inventory position: A policy-defined measure including relevant on-hand, scheduled receipts, allocations, and backorders.

ABC analysis: Segmentation by annual consumption value.

XYZ analysis: Segmentation by demand predictability or variability.

Intermittent demand: Demand characterized by many zero periods and occasional nonzero requirements.

Multi-echelon inventory: Inventory distributed across interconnected levels of a supply network.

Planner override: A documented human change to a system-generated forecast or recommendation.

Scenario analysis: Evaluation of alternative assumptions and their consequences.

Exception management: Identification and prioritization of conditions requiring attention.

Master data: Foundational information about items, locations, suppliers, calendars, and relationships.

Forecast value add: The measured contribution of an intervention compared with a baseline forecast.

Review Questions

5 Short Answer Questions

1. Why are shipments not always a valid measure of final demand?
2. What is the difference between forecast error and forecast bias?
3. What is the purpose of ABC analysis?
4. Define cycle service level and fill rate.
5. Why should planner overrides be recorded and evaluated?

5 Long Answer Questions

1. Explain how demand-planning software connects forecasting, inventory policy, replenishment, and execution.
2. Compare moving-average, exponential-smoothing, causal, and machine-learning forecasting approaches, including their assumptions and limitations.
3. Explain how demand variability, lead-time variability, and service targets influence safety stock.
4. Discuss the advantages of multi-echelon inventory planning over independent local optimization.
5. Evaluate the governance, ethical, and data-quality requirements for implementing AI-supported demand planning.

3 Analytical / Application Questions

1. A product has average weekly demand of 80 units, weekly demand standard deviation of 15 units, and a fixed lead time of 4 weeks. Using a 95% cycle-service target with z = 1.645, calculate expected lead-time demand, safety stock, and reorder point. State the assumptions required.
2. A planner increases a statistical forecast by 20% for six consecutive months, but actual demand remains close to the baseline forecast. Design a method to determine whether the overrides added forecast value and recommend a governance response.
3. A company has one national distribution center and six regional facilities. Each facility independently holds safety stock for the same customers. Propose an analytical approach to determine whether multi-echelon repositioning could reduce total inventory without reducing end-customer service.

Multiple Choice Questions with Answers

Question 1

Under the convention e_t = A_t − F_t, what does a positive forecast error indicate?

A. Actual demand exceeded the forecast

B. The forecast exceeded actual demand

C. The forecast had zero bias

D. The item had no demand

Correct Answer: A. Actual demand exceeded the forecast

Question 2

Which measure is especially problematic when actual demand is zero?

A. MAE

B. ME

C. MAPE

D. RSFE

Correct Answer: C. MAPE

Question 3

A product has average demand of 50 units per week and a fixed lead time of 3 weeks. If safety stock is 40 units, what is the reorder point?

A. 90 units

B. 150 units

C. 190 units

D. 200 units

Correct Answer: C. 190 units

Question 4

What is the principal purpose of safety stock?

A. To eliminate all inventory carrying cost

B. To protect against uncertainty in demand or replenishment

C. To replace demand forecasting

D. To increase supplier lead time

Correct Answer: B. To protect against uncertainty in demand or replenishment

Question 5

Which statement best distinguishes cycle service level from fill rate?

A. They are always mathematically identical

B. Cycle service level concerns avoiding a stockout during a cycle, while fill rate concerns the proportion of demand immediately fulfilled

C. Fill rate applies only to suppliers

D. Cycle service level measures forecast bias

Correct Answer: B. Cycle service level concerns avoiding a stockout during a cycle, while fill rate concerns the proportion of demand immediately fulfilled

Question 6

Which segmentation dimension classifies items according to annual consumption value?

A. ABC

B. XYZ

C. Criticality-only analysis

D. Seasonality indexing

Correct Answer: A. ABC

Question 7

Why can independent optimization of each facility create excessive total inventory?

A. It prevents any facility from holding stock

B. It may duplicate protection against the same network uncertainty

C. It always reduces service levels

D. It removes supplier constraints

Correct Answer: B. It may duplicate protection against the same network uncertainty

Question 8

Which practice provides the strongest auditability for a planner override?

A. Deleting the original forecast

B. Changing the forecast without recording a reason

C. Recording the original value, revised value, reason, author, and date

D. Allowing only sales personnel to make overrides

Correct Answer: C. Recording the original value, revised value, reason, author, and date

Question 9

In the simple exponential-smoothing equation F_t+1 = αA_t + (1 − α)F_t, what is the effect of increasing α?

A. The forecast responds more quickly to recent actual demand

B. The forecast ignores recent actual demand

C. The forecast becomes independent of prior forecasts

D. The forecast necessarily becomes seasonal

Correct Answer: A. The forecast responds more quickly to recent actual demand

Question 10

What is the most appropriate description of a planning scenario?

A. A guaranteed prediction of future demand

B. A controlled alternative set of assumptions used to evaluate consequences

C. A replacement for master data

D. A permanent manual override

Correct Answer: B. A controlled alternative set of assumptions used to evaluate consequences