Inventory Data, Costs, KPIs and Segmentation

Introduction

Inventory control depends on the quality of the information used to make decisions. A technically sound ordering model can produce poor results when stock records are inaccurate, costs are incomplete, demand is misclassified, or obsolete items remain hidden in aggregate totals. Inventory data therefore has two roles: it describes the physical system and it provides the economic basis for policy decisions.

For production managers, inventory is not simply material stored in a warehouse. It includes raw materials, components, work-in-process, maintenance supplies, finished goods, packaging, goods in transit, and sometimes consignment stock. Each category has different operational risks, financial implications, and control requirements.

This chapter develops a framework for assessing inventory costs, measuring performance, validating inventory records, and segmenting items for differentiated control. It combines financial, operational, and criticality perspectives. The objective is not to identify one universally superior classification, but to match control intensity with economic value, uncertainty, movement, and operational consequence.

Learning objectives

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

1. Distinguish ordering, setup, holding, shortage, and obsolescence costs.
2. Calculate inventory turnover, days of supply, inventory investment, and related KPIs.
3. Evaluate inventory-record accuracy using quantitative and process measures.
4. Apply ABC, XYZ, FSN, and VED classification methods.
5. Design a multi-criteria segmentation scheme for production inventories.
6. Interpret KPI results for managerial action while recognising their limitations.

1. Inventory data as a decision resource

Inventory decisions require both master data and transaction data. Master data describe relatively stable attributes: item code, description, unit of measure, supplier, lead time, standard cost, order multiple, shelf life, storage conditions, and criticality. Transaction data record receipts, issues, transfers, returns, adjustments, purchase orders, production consumption, and cycle-count results.

A credible inventory information system should establish one unique item identifier, a controlled unit of measure, effective dates for engineering changes, traceability of adjustments, and role-based approval. Duplicate item records are particularly damaging. For example, two codes for the same bearing can divide demand, distort reorder calculations, and cause unnecessary purchasing.

Important data-quality dimensions include:

• Completeness: required fields are populated.
• Accuracy: recorded values correspond to physical and commercial reality.
• Consistency: units, naming conventions, and classifications agree across systems.
• Timeliness: transactions are recorded close to the event.
• Validity: values satisfy business rules, such as positive lead time or an approved supplier.
• Uniqueness: one physical and commercial item is not represented by uncontrolled duplicate records.

Inventory data should also distinguish quantity states. On-hand stock is not automatically available stock. A useful operational identity is:

The terms must be measured in the same unit and at the same time. “Confirmed receipts” should not include speculative or cancelled orders. The identity is a planning convention, not a substitute for physical verification.

2. Inventory cost architecture

Inventory costs are the economic consequences of acquiring, storing, using, or failing to provide material. Cost classification must be consistent with the decision being analysed. A cost that is fixed for one decision may be variable for another.

2.1 Purchase and acquisition cost

Purchase cost is the supplier price multiplied by quantity, adjusted for discounts, freight, duties, taxes that are not recoverable, and other directly attributable acquisition costs. If a unit costs $18 and inbound freight is $2 per unit, the relevant acquisition cost is $20 per unit, unless accounting policy treats freight differently.

Purchase cost is distinct from ordering cost. Purchase cost changes with the number of units acquired; ordering cost is generally associated with initiating and processing an order.

2.2 Ordering cost

Ordering cost includes purchase requisition processing, supplier communication, approval, receiving, inspection, invoice matching, and payment administration. It may be estimated as:

where D is annual demand in units per year, Q is order quantity in units per order, and S is ordering cost per order. The formula assumes relatively stable demand, complete replenishment, and a constant cost per order. It is most useful when order-processing effort is material and varies with order frequency.

2.3 Setup cost

Setup cost is the cost of preparing production equipment for a run. It may include cleaning, tooling, fixture changes, programming, first-piece inspection, line clearance, and lost capacity during setup. In a manufacturing environment, setup cost is economically analogous to ordering cost, but it is incurred internally and affects production lot size.

A setup cost should not be confused with total conversion cost. Only the incremental cost caused by initiating a production run belongs in a lot-sizing calculation. If a setup requires three employees for two hours at $30 per employee-hour, the direct labour component is:

Additional tooling, scrap, and downtime costs should be included when they are decision-relevant.

2.4 Holding or carrying cost

Holding cost is the cost of possessing inventory over time. Components commonly include:

• capital committed to inventory;
• warehouse space, utilities, and handling;
• insurance and property-related charges;
• deterioration, damage, shrinkage, and pilferage;
• inventory administration and cycle counting;
• obsolescence risk.

A common approximation is:

where C is the value per unit and i is the annual holding-rate fraction. If unit value is $40 and the annual holding rate is 22%, annual holding cost is $8.80 per unit-year. The rate must not mix percentages and decimals: 22% = 0.22.

For average inventory Ī:

The holding rate is not a natural constant. It is a managerial estimate and should be reviewed when interest rates, warehouse costs, insurance, damage rates, or obsolescence exposure change.

2.5 Shortage cost

Shortage cost arises when required inventory is unavailable. It may include emergency freight, expediting, production interruption, lost contribution margin, customer penalties, lost goodwill, line-restart costs, and the administrative burden of recovery. Shortage cost is often difficult to estimate because the consequences are nonlinear: one missing low-value component can stop a high-value assembly line.

Shortage cost may be classified as:

• backorder cost, when demand is delayed but eventually fulfilled;
• lost-sales cost, when the customer or internal user obtains no replacement;
• production disruption cost, when a shortage interrupts operations;
• service and reputation cost, which may be long-term and difficult to quantify.

A sound analysis should distinguish avoidable shortage cost from unavoidable demand variation and should avoid double counting lost margin and penalty costs.

2.6 Obsolescence cost

Obsolescence occurs when inventory cannot be used or sold at its recorded value because of product discontinuation, engineering change, technology substitution, regulatory change, expiry, fashion, or declining demand. Obsolescence differs from ordinary holding cost because it can destroy a large proportion of inventory value in a single event.

A simplified expected-obsolescence estimate is:

This is an expected-value calculation, not a guaranteed loss. It requires a defensible estimate of the quantity exposed and the probability and severity of loss. Where recoverable value exists, expected recoverable value should be subtracted from gross value at risk.

2.7 Total relevant inventory cost

For a particular policy, relevant annual cost can be represented as:

The expression is a framework rather than a universal accounting identity. Some components may be excluded if they do not change between alternatives. For example, if unit purchase price is identical for two policies, purchase cost may be omitted from the comparison, although it remains part of total expenditure.

3. Inventory performance indicators

KPIs should connect inventory investment with service, flow, accuracy, and risk. No single KPI is sufficient. A high turnover ratio can indicate efficient flow, but it can also result from understocking and frequent shortages.

3.1 Inventory investment

Inventory investment is the monetary value tied up in inventory:

The valuation cost must be defined. It may be standard cost, weighted-average cost, FIFO cost, replacement cost, or another approved basis. Managerial reports should not silently compare values calculated under different bases.

3.2 Inventory turnover

Inventory turnover measures how many times average inventory is consumed or sold during a period:

The numerator and denominator should be comparable. Cost of goods sold should normally be compared with inventory valued at cost, not selling price. Average inventory may be calculated as beginning inventory plus ending inventory divided by two, but monthly or daily averages are preferable when inventory is seasonal.

A worked example illustrates the calculation. Suppose annual material usage at cost is $12 million. Beginning inventory is $2.4 million and ending inventory is $1.6 million.

The result means that the average inventory value was consumed six times during the year; it does not mean that every item moved six times.

3.3 Days of supply and days inventory outstanding

Days of supply estimates how long current inventory would last at a stated usage rate:

If using value:

With $2 million average inventory, $12 million annual usage, and 300 operating days:

Days of supply is meaningful only when the demand basis, operating-day convention, and inventory scope are specified. It can be misleading for slow-moving, project-specific, or highly seasonal items.

The relationship between turnover and days of supply, under consistent annual usage and average inventory definitions, is:

Thus, six turns and 300 operating days imply 50 days. This relationship is not valid when the two measures use different scopes or valuation bases.

3.4 Service and shortage indicators

Common service indicators include fill rate, cycle service level, backorder rate, stockout frequency, and line-stop incidents. They measure different phenomena:

• Fill rate is the proportion of demand quantity supplied immediately.
• Cycle service level is the proportion of replenishment cycles completed without a stockout.
• Backorder rate measures demand delayed rather than lost.
• Line-stop frequency focuses on production continuity.

A business should define the numerator, denominator, time window, and treatment of partial deliveries. A fill rate based on order lines may give a different result from one based on units.

3.5 Record accuracy

Record accuracy compares system quantity with physically verified quantity. A simple item-level measure is:

If 920 of 1,000 counted records are within the approved tolerance, accuracy is 92%.

A value-based measure is:

This measure gives more weight to expensive discrepancies. The absolute difference prevents positive and negative errors from cancelling. Tolerance rules must be declared; otherwise a record can be labelled accurate despite a commercially important variance.

Record accuracy should be supplemented by root-cause indicators: unposted receipts, picking errors, unit-of-measure conversions, location errors, unauthorised adjustments, counting delays, and inaccurate bills of material.

4. Inventory segmentation

Segmentation allocates managerial attention according to risk and decision relevance. It should not be treated as a permanent label. Demand, value, criticality, and movement can change over time.

4.1 ABC analysis

ABC analysis ranks items by annual usage value:

Items are sorted from highest to lowest annual usage value, and cumulative percentages are calculated. A common policy is A items as approximately the top 70–80% of value, B items as the next 15–25%, and C items as the remaining 5–10%. These thresholds are conventions, not laws.

ABC controls often include:

ABC ignores criticality. A low-value safety relay may be classified C economically but still require A-level control from a production-continuity perspective.

4.2 XYZ analysis

XYZ analysis classifies demand predictability. One common method uses the coefficient of variation:

where σ_D is the standard deviation of periodic demand and μ_D is mean periodic demand. Both must use the same time unit. Typical thresholds might classify X as low variability, Y as moderate variability, and Z as high variability, but the thresholds must reflect the organisation’s demand pattern.

For intermittent demand, a coefficient of variation alone may be inadequate because many zero-demand periods can distort interpretation. Demand interval, intermittency, and forecast error should be considered.

4.3 FSN analysis

FSN classifies items by movement:

• Fast-moving: frequent issues or high recent usage;
• Slow-moving: occasional issues or low usage;
• Non-moving: no issue during a defined observation period.

The observation window must be stated. A spare part with no issue for twelve months may be essential for a machine that fails only once every two years. FSN is therefore a movement indicator, not a disposal decision.

4.4 VED analysis

VED classifies items by operational or functional criticality:

• Vital: absence can stop production, compromise safety, or create severe service consequences;
• Essential: absence causes substantial disruption but alternatives or short-term workarounds exist;
• Desirable: absence has limited operational consequence.

Criticality assessment should involve maintenance, production, engineering, quality, safety, and finance. It should consider redundancy, repair lead time, substitution, failure consequence, and regulatory requirements.

4.5 Combining classifications

A robust segmentation uses multiple dimensions. For example, an item may be AXV: high annual value, predictable demand, and vital criticality. Another may be CZV: low annual value, erratic demand, and vital criticality. The second item should not receive casual control merely because its annual value is low.

A practical multi-criteria process is:

1. Define the decision purpose: counting, replenishment, sourcing, or risk review.
2. Select dimensions relevant to that purpose.
3. Standardise data definitions and observation periods.
4. Assign classifications using transparent thresholds.
5. Review exceptional combinations, especially high criticality and high uncertainty.
6. Translate segments into explicit policies.
7. Reassess periodically and after major engineering or supply changes.

A weighted score may be used when governance is strong:

where V is normalised value, U is uncertainty, K is criticality, R is supply risk, and the weights satisfy w₁ + w₂ + w₃ + w₄ = 1. Normalisation is essential because dollars, percentages, and ordinal ratings cannot be added directly. Weights should be approved by stakeholders and tested through sensitivity analysis. A score must not conceal a mandatory safety or regulatory requirement.

5. Worked integrated example

A plant manages four purchased components. Annual demand, unit cost, demand variability, and criticality are shown below.

Total annual usage value is $690,000. Item P represents 72.46%, Q 23.19%, R 2.90%, and S 1.45%. Ranked cumulatively, P and Q account for 95.65% of value. Under a locally defined 70–95–100 rule, P is A, Q is B at the boundary depending on policy, and R and S are C. This illustrates why thresholds must be stated rather than assumed.

Demand variability suggests P and S are X, Q is Y, and R is Z if the organisation uses thresholds X ≤ 0.20, 0.20 < Y ≤ 0.50, and Z > 0.50. Item R is economically C but operationally vital and highly erratic. It should receive a critical-spares policy, not an automatic low-control policy. Possible actions include verified equipment bills of material, approved substitutes, failure-mode review, and a service-level decision based on downtime cost.

6. Applied case: illustrative precision-pump manufacturer

Background

Precision Pumps Ltd manufactures industrial dosing systems. Its enterprise system reports annual inventory turnover of 8.2, apparently better than the target of 7.0. However, production supervisors report frequent emergency purchases and line interruptions.

Problem

An audit identifies three symptoms: 88% location accuracy, duplicate item codes for seals, and a high proportion of non-moving maintenance inventory. The turnover KPI excludes some maintenance stores and uses ending inventory rather than average inventory.

Analysis

The supply-chain team first reconciles item master records and standardises units. It then recounts A items and all vital spares. Inventory investment is recalculated using a consistent standard-cost basis. After adjustment, annual usage value is $24 million and average inventory is $3.6 million:

The apparent 8.2 turns had resulted from an incomplete denominator. A segmentation exercise identifies high-value predictable production materials, low-value vital seals, and slow-moving obsolete tooling. The team also separates usable, quality-hold, and blocked stock in the availability report.

Approach and solution

The company introduces the following controls:

• cycle counts weekly for A items and vital spares;
• barcode scanning at receipt, issue, and location transfer;
• approval for item-master creation and duplicate-code merging;
• quarterly review of Z-demand and non-moving items;
• engineering sign-off before disposal of maintenance spares;
• a dashboard showing turnover together with fill rate, line stops, record accuracy, and aged inventory value.

Outcome

After six months, location accuracy rises to 98%, emergency purchases decline, and reported turnover falls initially because previously excluded inventory is included. This fall is not operational deterioration; it is measurement correction. Subsequent improvement is assessed against the corrected baseline.

Learning points

The case demonstrates that KPIs are only as reliable as their definitions and scope. Higher turnover is not automatically better, and disposal of non-moving stock requires technical and safety review. Data governance, physical control, and segmentation must operate together.

7. Limitations, governance, and contemporary practice

Inventory metrics can create dysfunctional behaviour. Managers may reduce inventory just before period-end, delay receipts, split orders, or avoid stocking low-value critical items to improve a narrow KPI. Balanced performance systems should therefore combine cost, service, accuracy, risk, and working-capital indicators.

Data governance requires segregation of duties. The person who creates an item or approves an adjustment should not be the sole person who verifies the physical stock. Audit trails should preserve the user, time, reason, and authorisation for every material adjustment. Safety-critical and regulated items require traceability, expiry control, and documented disposition.

Digital technologies can improve control when process discipline exists. Barcode and RFID systems reduce transaction latency; warehouse-management systems improve location visibility; sensors can monitor temperature and shelf life; and analytical tools can detect unusual adjustments or duplicate descriptions. AI may assist demand classification, anomaly detection, and obsolescence-risk prediction, but its outputs require explainable inputs, human review, bias monitoring, and protection of commercially sensitive data. An algorithm trained on inaccurate transactions can automate error at scale.

Best practice is to:

1. define every KPI mathematically and operationally;
2. reconcile physical, financial, and system records;
3. use consistent valuation and time bases;
4. combine value, variability, movement, and criticality;
5. assign each segment a specific control policy;
6. investigate exceptions rather than relying only on averages;
7. review thresholds after changes in product, supply, technology, or risk;
8. measure service and operational continuity alongside inventory reduction.

Chapter Summary

Inventory control begins with trustworthy data. Master-data quality, transaction discipline, unit consistency, and physical verification determine whether analytical models are credible. Relevant inventory costs include acquisition, ordering, setup, holding, shortage, and obsolescence costs; their relevance depends on the decision and the assumptions used.

Inventory turnover relates annual usage or cost of goods sold to average inventory value. Days of supply expresses inventory relative to average daily demand, while inventory investment measures capital committed to stock. These indicators must use consistent scope, valuation, and time bases. Record accuracy should be measured both by item count and by value, supported by root-cause analysis.

ABC highlights economic value, XYZ highlights demand variability, FSN highlights movement, and VED highlights operational criticality. Their combination provides a stronger basis for differentiated control than any single classification. Multi-criteria scoring can support prioritisation, but normalisation, approved weights, sensitivity testing, and protection of safety requirements are essential. Finally, KPIs must be governed carefully because narrow targets can produce unintended behaviour.

Key Terms

Inventory investment: Monetary value committed to inventory.

Ordering cost: Cost incurred to place and process a replenishment order.

Setup cost: Cost of preparing production resources for a manufacturing run.

Holding cost: Cost of possessing inventory over time.

Shortage cost: Economic consequence of unavailable inventory.

Obsolescence: Loss of usable or saleable value due to ageing, change, or expiry.

Inventory turnover: Annual usage or cost of goods sold divided by average inventory value.

Days of supply: Inventory quantity or value divided by average daily demand or usage.

Record accuracy: Degree to which system records agree with physical stock within tolerance.

ABC analysis: Classification by annual usage value.

XYZ analysis: Classification by demand variability or predictability.

FSN analysis: Classification by movement frequency or recency.

VED analysis: Classification by vital, essential, and desirable criticality.

Annual usage value: Annual demand multiplied by unit cost.

Coefficient of variation: Standard deviation divided by mean demand.

Fill rate: Proportion of demand supplied immediately.

Cycle service level: Proportion of replenishment cycles without a stockout.

Master data: Controlled descriptive information about inventory items.

Inventory segmentation: Grouping items to apply differentiated controls.

Data governance: Policies and responsibilities that ensure reliable, secure, and accountable data.

Review Questions

Short Answer Questions

1. What is the difference between ordering cost and setup cost?
2. Why should turnover use average inventory rather than ending inventory when possible?
3. Define inventory record accuracy and state one limitation of a count-based measure.
4. What does the Z category represent in XYZ analysis?
5. Why may a C-class item require strict control?

Long Answer Questions

1. Explain the main components of holding cost and discuss why the holding-rate estimate should be periodically reviewed.
2. Compare ABC, XYZ, FSN, and VED classification, including the decision dimension addressed by each.
3. Discuss the assumptions and limitations of inventory turnover and days-of-supply measures.
4. Develop a governance framework for improving inventory-record accuracy in a manufacturing plant.
5. Explain how shortage and obsolescence costs should influence inventory policy even when annual usage value is low.

Analytical / Application Questions

1. A plant has annual usage at cost of $9 million, beginning inventory of $1.8 million, and ending inventory of $2.2 million. Calculate average inventory, turnover, and days of supply using 300 operating days.
2. An item has annual demand of 1,200 units, unit cost of $35, and demand standard deviation of 30 units per month with mean monthly demand of 100 units. Calculate annual usage value and the coefficient of variation, and explain the classification implications.
3. Design a multi-criteria segmentation policy for a low-value spare part that is highly intermittent, vital to safety, and supplied with a nine-month lead time. State the data required, control actions, and risks of relying on ABC alone.

Multiple Choice Questions with Answers

Question 1

A plant uses $12 million of annual material usage at cost. Beginning inventory is $2.4 million and ending inventory is $1.6 million. What is inventory turnover?

A. 4 turns per year

B. 6 turns per year

C. 7.5 turns per year

D. 12 turns per year

Correct Answer: B. 6 turns per year

Question 2

Which cost is most directly associated with preparing a machine for a new production run?

A. Holding cost

B. Shortage cost

C. Setup cost

D. Obsolescence cost

Correct Answer: C. Setup cost

Question 3

An item costs $40 and has an annual holding rate of 22%. What is its approximate annual holding cost per unit?

A. $0.88

B. $8.80

C. $18.18

D. $62.00

Correct Answer: B. $8.80

Question 4

Which classification primarily reflects operational criticality?

A. ABC

B. XYZ

C. FSN

D. VED

Correct Answer: D. VED

Question 5

Annual usage value is calculated as:

A. Unit cost ÷ annual demand

B. Annual demand × unit cost

C. Average inventory ÷ annual demand

D. Annual demand × holding rate

Correct Answer: B. Annual demand × unit cost

Question 6

If 920 of 1,000 counted records are within the approved tolerance, count-based record accuracy is:

A. 8%

B. 9.2%

C. 92%

D. 108.7%

Correct Answer: C. 92%

Question 7

An item has mean monthly demand of 100 units and standard deviation of 30 units. Its coefficient of variation is:

A. 0.03

B. 0.30

C. 3.00

D. 30.00

Correct Answer: B. 0.30

Question 8

Which statement about a C-class item is most accurate?

A. It is always operationally unimportant.

B. It must have highly predictable demand.

C. It has low annual usage value but may still be critical.

D. It has no inventory carrying cost.

Correct Answer: C. It has low annual usage value but may still be critical.

Question 9

Why are absolute differences used in the value-accuracy formula?

A. To make all items equal in value

B. To prevent positive and negative discrepancies from cancelling

C. To exclude expensive items

D. To convert quantity into demand

Correct Answer: B. To prevent positive and negative discrepancies from cancelling

Question 10

Which practice best supports reliable multi-criteria segmentation?

A. Adding raw dollars, percentages, and ratings without transformation

B. Using one permanent classification for every decision

C. Normalising measures and reviewing exceptional criticality cases

D. Classifying all slow-moving items as obsolete

Correct Answer: C. Normalising measures and reviewing exceptional criticality cases