Performance Measurement, Costing and Analytics

Introduction

Supply chains are managed through decisions, but decisions require evidence. Performance measurement converts operational activity into information that managers can use to control cost, improve service, protect cash, reduce risk and support sustainability. A warehouse may report high productivity while customers experience stockouts; a purchasing department may obtain low unit prices while increasing inventory and obsolescence. Effective measurement therefore requires a balanced view of the whole supply chain rather than isolated departmental statistics.

This chapter introduces practical performance measurement, supply-chain costing and basic analytics. It focuses on six performance dimensions: cost, cash, quality, delivery, flexibility and sustainability. It also explains inventory turns, fill rate, on-time-in-full (OTIF), cash-to-cash cycle time and forecast accuracy. The emphasis is on interpretation: a KPI is useful only when its definition, data, target and managerial response are clear.

Learning objectives

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

• explain the purpose and characteristics of effective supply-chain KPIs;
• distinguish leading and lagging indicators and design a balanced dashboard;
• calculate and interpret inventory turns, fill rate, OTIF, cash-to-cash cycle time and forecast accuracy;
• classify supply-chain costs and identify trade-offs between cost and service;
• apply descriptive and basic diagnostic analytics to operational decisions;
• evaluate the limitations, governance requirements and ethical risks of performance data; and
• recommend practical improvement actions from a performance dashboard.

1. The role of performance measurement

Performance measurement is the systematic collection, calculation and interpretation of evidence about organisational results. A metric is a defined measure, such as units shipped per day. A key performance indicator (KPI) is a metric judged to be strategically important and monitored against a target, threshold or benchmark.

Measurement serves four related purposes:

1. Visibility: it shows what is happening across suppliers, facilities, transport and customers.
2. Control: it identifies deviations that require corrective action.
3. Learning: it helps managers understand whether process changes produce improvement.
4. Alignment: it connects operational behaviour with organisational objectives.

A good KPI has a precise definition, an accountable owner, a reliable data source, an appropriate frequency, a target and a specified response. The question “What is our delivery performance?” is incomplete unless delivery, order population, time period and treatment of partial shipments are defined.

Leading and lagging indicators

A lagging indicator measures an outcome after it occurs. Examples include total logistics cost, customer complaints and annual carbon emissions. A leading indicator signals conditions likely to influence future outcomes, such as supplier schedule adherence, forecast bias, preventive-maintenance completion or the percentage of orders released with complete information.

Lagging indicators are important for accountability, but leading indicators support earlier intervention. A dashboard should contain both. For example, poor OTIF is a lagging result; rising supplier confirmation delays may be a leading warning.

Balanced measurement

A narrow focus on one objective creates dysfunctional behaviour. If a warehouse is rewarded only for low labour cost, it may reduce staffing and increase picking errors. If purchasing is rewarded only for price reduction, it may buy excessive quantities or accept poor quality. A balanced dashboard connects financial, customer, process, people and sustainability outcomes.

2. Designing a KPI system

KPI design should begin with strategic and customer requirements, then proceed through process mapping and data definition. A practical design sequence is:

1. identify the decision or objective;
2. define the process and performance dimension;
3. specify numerator, denominator, exclusions and time period;
4. identify the data owner and source system;
5. establish a baseline and realistic target;
6. assign review frequency and escalation rules; and
7. test whether the KPI encourages the desired behaviour.

For example, a fill-rate KPI should state whether it is measured by order lines, units or sales value. These versions can produce different results. A line fill rate may be high when important high-volume items are unavailable, while a unit fill rate may better represent physical customer demand.

Targets, benchmarks and thresholds

A target is the desired level of performance. A benchmark is a comparison with another period, facility, supplier or recognised standard. A threshold identifies a condition requiring investigation. Targets should account for capability, customer requirements, cost and risk. An unrealistic target can encourage data manipulation or unsafe work practices.

A KPI should also be segmented. Overall OTIF can conceal poor performance for a major customer, product family, region or carrier. Useful dimensions include supplier, SKU, route, shift, facility, customer and order priority.

3. Core supply-chain KPIs and calculations

The following formulas are common management measures. Organisations must document their own conventions, especially for cost of goods sold, inventory valuation, order dates and partial fulfilment.

3.1 Inventory turns

Inventory turns estimate how many times average inventory is consumed and replenished during a period.

Average inventory is commonly calculated as:

Cost of goods sold and inventory must use the same valuation basis and period. The result is expressed as turns per period, usually turns per year. High turns may indicate efficient inventory use, but excessively high turns can indicate insufficient safety stock and greater stockout risk. Turns should therefore be interpreted with service and availability measures.

Worked example. Annual cost of goods sold is $2,400,000. Opening inventory is $420,000 and closing inventory is $380,000.

Average inventory = ($420,000 + $380,000) ÷ 2 = $400,000.

Inventory turns = $2,400,000 ÷ $400,000 = 6 turns per year.

An approximate inventory holding period can be calculated as:

Using 365 days, the example gives 60.8 days. This approximation assumes relatively stable consumption and should not be treated as an exact age measure.

3.2 Fill rate

Fill rate measures the proportion of demand supplied immediately from available stock. A unit fill rate is:

The numerator must relate to the same demand population as the denominator. A line fill rate instead uses order lines supplied completely divided by order lines demanded. Fill rate is not identical to OTIF: an order may be filled from stock but delivered late, or delivered on time with a missing line.

Example. A distributor receives demand for 10,000 units and supplies 9,400 immediately. Fill rate = 9,400 ÷ 10,000 × 100 = 94%.

3.3 On-time-in-full (OTIF)

OTIF measures whether an order meets both the agreed date and the agreed quantity. At order level:

An order counts as successful only if both conditions are satisfied. If 930 of 1,000 due orders meet both conditions, OTIF is 93%. Separate on-time and in-full measures may be used diagnostically, but their simple average is not generally equivalent to OTIF.

Definitions must specify the promised date, delivery timestamp, tolerance window, quantity tolerance, treatment of customer-requested changes and whether cancelled orders are excluded.

3.4 Forecast accuracy

Forecast accuracy compares a forecast with actual demand. One accessible measure is absolute percentage error (APE):

Mean absolute percentage error (MAPE) is the arithmetic mean of APE values:

Here, t identifies a time period and n is the number of periods. MAPE is unsuitable or unstable when actual demand is zero or very close to zero. In such cases, use absolute error, scaled error or a measure based on total volume. Forecast bias should also be monitored:

Positive bias indicates systematic over-forecasting under this convention; negative bias indicates under-forecasting.

Example. Actual monthly demand is 1,000 units and forecast demand is 900. APE = |1,000 − 900| ÷ 1,000 × 100 = 10%. The forecast is low by 100 units, but its error is 10%.

3.5 Cash-to-cash cycle time

Cash-to-cash (C2C) cycle time estimates the number of days between paying suppliers and collecting cash from customers.

Where:

COGS means cost of goods sold, and “Days” is commonly 365 or 360, provided the same convention is used consistently. The measure is an approximation and depends on accounting policies and seasonality.

Example. Inventory days are 61, receivable days are 45 and payable days are 35. C2C = 61 + 45 − 35 = 71 days. Reducing C2C can release cash, but extending supplier payment without agreement may damage supplier viability or violate contractual terms.

4. Supply-chain costing

Supply-chain cost is broader than purchase price. Relevant cost categories include:

• purchase or manufacturing cost;
• inbound and outbound transport;
• warehousing, handling and storage;
• ordering, planning and information-system cost;
• quality inspection, warranty and returns;
• inventory carrying cost, including capital, insurance, damage and obsolescence;
• duties, taxes and compliance costs; and
• environmental and social costs where measured.

Total cost of ownership (TCO) evaluates the full economic effect of a sourcing decision rather than the invoice price alone. A simplified expression is:

The categories must be defined to avoid double counting. TCO is decision-specific: a supplier with a higher unit price may have lower defects, shorter lead time and lower safety-stock requirements.

Cost-to-serve

Cost-to-serve assigns supply-chain costs to customers, channels, products or orders. It can reveal that a high-revenue customer is unprofitable because of frequent small orders, urgent transport, extensive returns or special packaging. Activity-based costing may assign costs through cost drivers such as order lines, deliveries, pallets, kilometres or inspection hours.

Managers should distinguish fixed, variable, direct and indirect costs. For short-term decisions, relevant incremental costs may matter more than fully allocated costs. For long-term network design, capacity and fixed costs must also be considered.

5. Descriptive and diagnostic analytics

Descriptive analytics answers “What happened?” It includes totals, averages, percentages, trend lines, distributions and dashboards. Diagnostic analytics asks “Why did it happen?” It uses segmentation, comparison, drill-down, correlation analysis, process investigation and root-cause methods.

A practical analytics cycle is:

1. define the management question;
2. validate data completeness, accuracy and timeliness;
3. calculate the KPI consistently;
4. compare actual performance with target, baseline or peer group;
5. segment the result to locate concentration;
6. investigate plausible causes; and
7. select, implement and review an action.

A dashboard should show current value, target, trend, variance, owner and action status. Colour coding can help, but colour should not be the only signal because users may have visual impairments or print reports in monochrome.

Diagnostic tools

• Pareto analysis: ranks causes or categories to identify the few contributing most to a problem.
• Trend analysis: examines movement over time and detects seasonality or deterioration.
• Cohort or segment analysis: compares suppliers, products, customers, shifts or routes.
• Drill-down: moves from an aggregate KPI to transactions or process steps.
• Five-whys analysis: repeatedly asks why a problem occurred until a controllable underlying cause is identified.
• Fishbone analysis: groups possible causes under categories such as people, process, technology, materials and environment.

Correlation is not proof of causation. For example, higher transport cost may coincide with poor OTIF because urgent shipments are used to recover late orders; the cost may be a consequence rather than the original cause.

Data quality and governance

Reliable analytics requires common definitions, master-data control, timestamps, version history and access controls. Important governance practices include a KPI dictionary, named data owners, documented calculation rules, audit trails and reconciliation with financial records. Personal data should be minimised and protected. Employee-level dashboards can become unethical surveillance if used without legitimate purpose, transparency or due process.

6. Flexibility and sustainability measurement

Flexibility is the ability to respond to changes in volume, mix, timing, supply or customer requirements. Possible measures include time to increase output, percentage of demand accommodated within a defined period, changeover time and the proportion of capacity available for priority orders. Flexibility has a cost; excess capacity may reduce utilisation, while no spare capacity increases disruption risk.

Sustainability metrics should be connected to operational decisions. Examples include greenhouse-gas emissions per tonne-kilometre, energy per unit, waste diverted from landfill, water use, injury frequency and supplier corrective-action closure. Measurement boundaries must be stated—for example, whether transport emissions include subcontractors and whether emissions are calculated from fuel data or estimated factors.

A low emissions-per-unit result may arise from improved efficiency, but it can also be distorted by increased load utilisation, changes in product mix or exclusion of upstream activities. Sustainability dashboards should therefore combine intensity measures with absolute totals and should avoid unsupported claims.

7. Applied case: Northstar Homeware (illustrative)

Background

Northstar Homeware is an illustrative regional distributor of kitchen equipment. Its management dashboard reports annual COGS of $12 million, opening inventory of $2.4 million and closing inventory of $2.8 million. During the latest quarter, 4,800 orders were due; 4,320 were delivered on time and in full. Demand was 250,000 units, of which 232,500 were supplied immediately. Inventory days are 79, receivable days are 42 and payable days are 31.

Problem

Managers believe that purchasing has improved because supplier prices fell by 3%. However, customer complaints and expedited freight have increased. The executive team must determine whether the apparent purchasing improvement reflects total supply-chain improvement.

Analysis

Average inventory = ($2.4m + $2.8m) ÷ 2 = $2.6m.

Inventory turns = $12m ÷ $2.6m ≈ 4.62 turns per year.

OTIF = 4,320 ÷ 4,800 × 100 = 90%.

Unit fill rate = 232,500 ÷ 250,000 × 100 = 93%.

C2C = 79 + 42 − 31 = 90 days.

The results show weak service and a substantial cash commitment. The difference between the 93% fill rate and 90% OTIF suggests that some demand is available from stock but is not delivered within the agreed time, although the measures cover different periods and populations and should not be assumed directly comparable.

A diagnostic drill-down identifies three patterns: a small group of imported items has long and variable lead times; promotional forecasts are consistently optimistic; and urgent replenishment shipments are concentrated on those items. The 3% purchase-price reduction is therefore accompanied by higher inventory, premium freight and lost sales risk.

Approach and solution

Northstar introduces a balanced supplier dashboard containing price variance, defect rate, confirmed lead-time reliability, OTIF, fill rate and emissions per shipment. It segments SKUs by demand value and volatility, establishes separate service policies, and requires promotional forecasts to be reviewed by sales, marketing and supply-chain staff. For volatile imported items, it evaluates TCO rather than unit price and sets reorder parameters using updated lead-time data. The company also creates a weekly exception review for late supplier confirmations and stockout risk.

Outcome

After three review cycles, management expects to evaluate whether OTIF improves without excessive inventory growth, whether premium freight declines and whether C2C decreases. The case does not claim a guaranteed numerical improvement; the important managerial point is that actions and outcomes must be measured over a defined period and checked for unintended effects.

Learning points

• Purchase price is not the same as total cost.
• Service KPIs must be defined at the same level before comparison.
• Aggregate results conceal SKU and supplier-level causes.
• A dashboard is valuable when it leads to accountable decisions.
• Improvements should be evaluated across cost, cash, service and sustainability together.

8. Limitations and contemporary practice

KPIs can be gamed when rewards are tied too narrowly to numerical targets. Examples include delaying order entry, redefining promised dates, excluding difficult orders or transferring inventory between locations near period end. Controls include independent audits, balanced measures, exception review and qualitative management judgement.

Seasonality, product mix, inflation, exchange rates and unusual disruptions can make period-to-period comparisons misleading. Managers should use like-for-like comparisons and explain contextual changes. Statistical significance is not required for every operational decision, but large investments should use stronger evidence than a simple visual trend.

Modern supply-chain platforms combine enterprise-resource-planning data, warehouse systems, transport-management systems, supplier portals and external data. Artificial intelligence can assist anomaly detection, demand sensing and natural-language dashboard queries. It does not remove the need for sound definitions, representative data, human oversight and explainable decisions. Automated recommendations should be tested for bias, data leakage, security weakness and inappropriate optimisation of one KPI at the expense of the wider system.

Best-practice checklist

• Begin with customer and strategic requirements.
• Define every KPI, including numerator, denominator, scope and exclusions.
• Use balanced measures across cost, cash, quality, delivery, flexibility and sustainability.
• Combine leading indicators with lagging outcomes.
• Segment results before diagnosing causes.
• Reconcile operational data with financial records.
• Assign an owner, target, review frequency and escalation rule.
• Investigate trade-offs rather than celebrating isolated improvements.
• Protect personal and commercially sensitive data.
• Review KPI definitions as processes, technology and customer expectations change.

Chapter Summary

Supply-chain performance measurement transforms operational data into evidence for managerial action. Effective systems use clearly defined KPIs, balanced across cost, cash, quality, delivery, flexibility and sustainability. Inventory turns indicate the relationship between COGS and average inventory; fill rate measures immediate supply; OTIF requires both timeliness and completeness; forecast accuracy measures deviation from actual demand; and cash-to-cash cycle time links inventory, receivables and payables.

Cost analysis must extend beyond purchase price to logistics, quality, inventory, risk and compliance. Descriptive analytics explains what happened, while diagnostic analytics investigates where and why performance changed. Dashboards are most useful when they show trends, targets, segmentation, ownership and actions. Data governance, ethical use, consistent definitions and awareness of trade-offs are essential. The ultimate purpose of measurement is not reporting itself, but better decisions and sustained supply-chain performance.

Key Terms

Benchmark: A reference point used for comparison.

Cash-to-cash cycle time: Approximate time between paying suppliers and collecting customer cash.

Cost-to-serve: Cost of supplying a particular customer, product, channel or order.

Descriptive analytics: Analysis that summarises what happened.

Diagnostic analytics: Analysis that investigates why performance occurred.

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

Forecast accuracy: Degree to which forecast demand matches actual demand.

Inventory turns: Number of times average inventory is consumed during a period.

Key performance indicator: A strategically important, defined and monitored performance measure.

Leading indicator: Measure that signals conditions likely to affect future outcomes.

Lagging indicator: Measure of an outcome after it has occurred.

On-time-in-full: Percentage of due orders delivered by the agreed date and in the agreed quantity.

Pareto analysis: Ranking of causes or categories by contribution.

Performance dashboard: Visual display of selected measures, targets, trends and exceptions.

Total cost of ownership: Full cost associated with acquiring and using a product or service.

Forecast bias: Systematic tendency for forecasts to be above or below actual demand.

KPI dictionary: Controlled record of KPI definitions, sources, owners and calculation rules.

Review Questions

Short Answer Questions

1. What distinguishes a KPI from a general metric?
2. Why should a dashboard include both leading and lagging indicators?
3. Define inventory turns and state one limitation of interpreting it alone.
4. What two conditions must be satisfied for an order to count as OTIF?
5. Why is purchase price insufficient for a sourcing decision?

Long Answer Questions

1. Explain how a balanced KPI system can prevent dysfunctional departmental behaviour.
2. Describe the calculation and managerial interpretation of the cash-to-cash cycle time.
3. Compare descriptive and diagnostic analytics, giving supply-chain examples of each.
4. Discuss the data-governance requirements for a reliable performance dashboard.
5. Explain how sustainability measures can be distorted if their boundaries and denominators are poorly defined.

Analytical / Application Questions

1. A firm has annual COGS of $3,000,000, opening inventory of $500,000 and closing inventory of $700,000. Calculate average inventory and inventory turns. Interpret the result and identify one additional KPI required before recommending inventory reduction.
2. A company has 2,000 orders due, of which 1,760 are delivered on time and 1,840 are delivered in full. Only 1,650 satisfy both conditions. Calculate OTIF and explain why averaging the separate on-time and in-full percentages would not be the correct OTIF calculation.
3. A supplier offers a 4% price reduction but requires larger minimum order quantities. Propose a diagnostic analysis using at least four KPIs and explain the possible trade-offs.

Multiple Choice Questions with Answers

Question 1

A distributor has COGS of $2,400,000 and average inventory of $400,000. What are its inventory turns?

A. 4 turns

B. 6 turns

C. 8 turns

D. 10 turns

Correct Answer: B. 6 turns

Question 2

Which condition is required for an order to be counted as OTIF?

A. It must be delivered early

B. It must have the lowest transport cost

C. It must be delivered on time and in full

D. It must be supplied from local stock

Correct Answer: C. It must be delivered on time and in full

Question 3

Demand is 10,000 units and 9,400 units are supplied immediately. What is the unit fill rate?

A. 90%

B. 92%

C. 94%

D. 96%

Correct Answer: C. 94%

Question 4

Inventory days are 61, receivable days are 45 and payable days are 35. What is the cash-to-cash cycle time?

A. 71 days

B. 81 days

C. 91 days

D. 141 days

Correct Answer: A. 71 days

Question 5

Which measure is most directly a leading indicator?

A. Annual logistics cost

B. Customer complaints last quarter

C. Supplier confirmation delays

D. Last year’s carbon emissions

Correct Answer: C. Supplier confirmation delays

Question 6

What is the main purpose of total cost of ownership analysis?

A. To compare invoice prices only

B. To evaluate the wider costs of an acquisition or sourcing decision

C. To eliminate all fixed costs

D. To measure forecast bias

Correct Answer: B. To evaluate the wider costs of an acquisition or sourcing decision

Question 7

Which technique ranks causes or categories by their contribution to a problem?

A. Pareto analysis

B. Random sampling

C. Depreciation analysis

D. Capacity smoothing

Correct Answer: A. Pareto analysis

Question 8

Actual demand is 1,000 units and forecast demand is 900 units. What is the absolute percentage error?

A. 5%

B. 9%

C. 10%

D. 11%

Correct Answer: C. 10%

Question 9

Why should fill rate and OTIF not be treated as identical measures?

A. Fill rate never uses quantities

B. OTIF includes both timeliness and completeness

C. OTIF measures only inventory value

D. Fill rate is always a financial measure

Correct Answer: B. OTIF includes both timeliness and completeness

Question 10

Which practice best supports trustworthy KPI governance?

A. Changing definitions whenever results are unfavourable

B. Using undocumented manual adjustments

C. Maintaining a KPI dictionary and named data owners

D. Reporting only aggregate results

Correct Answer: C. Maintaining a KPI dictionary and named data owners