Inventory Economics, Performance and Segmentation

Introduction

Retail inventory is simultaneously an asset, a service mechanism, a source of risk and a representation of managerial judgement. A store or digital fulfilment operation must hold enough merchandise to satisfy demand, yet not so much that capital is trapped in slow-moving, damaged, seasonal or obsolete stock. Inventory management therefore involves economic trade-offs rather than the simple pursuit of maximum availability.

In retail, these trade-offs are intensified by short product life cycles, promotions, uncertain demand, omnichannel fulfilment, supplier variability and customer expectations of immediate service. A financially attractive inventory policy may produce poor availability; a service-oriented policy may generate excessive markdowns. Effective management integrates cost analysis, performance measurement and SKU segmentation.

This chapter develops that integration. It examines ordering, holding, shortage, markdown and obsolescence costs; explains the principal performance indicators used by retailers; and applies ABC, XYZ, FSN and multi-criteria segmentation. The emphasis is managerial: metrics are useful only when their definitions, denominators, time periods and decisions are aligned.

Learning objectives

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

1. Explain the economic trade-offs among ordering, holding, shortage, markdown and obsolescence costs.
2. Calculate and interpret inventory turnover, days of supply, sell-through, GMROI, stock-to-sales ratio, fill rate and availability.
3. Distinguish financial, operational and customer-service measures of inventory performance.
4. Apply ABC, XYZ and FSN classification and explain their limitations.
5. Construct a multi-criteria SKU segmentation for retail decision-making.
6. Recommend differentiated replenishment, review, markdown and audit policies.

1. The economics of retail inventory

1.1 Inventory as a portfolio of commitments

Inventory is purchased before the final customer decision is known. The retailer commits cash today in exchange for an uncertain future stream of sales, margin and service benefits. The economic value of a unit depends not only on its purchase cost but also on when it sells, whether it must be discounted, how much space it consumes, and whether it can be returned or redeployed.

The relevant cost categories are as follows.

These categories are interdependent. Increasing order quantity may reduce ordering cost per unit but increase average stock and therefore holding and obsolescence risk. Reducing stock may improve turnover but increase stockouts and lost sales. A cost model is consequently a decision model, not merely an accounting statement.

1.2 Ordering cost

Ordering cost is usually a fixed or semi-fixed cost per replenishment event. It may include purchase-order administration, supplier communication, receiving, inspection, invoice processing, transport setup and store labour. If annual demand is D units and the order quantity is Q units, the number of orders per year is D ÷ Q. If S is ordering cost per order, annual ordering cost is:

Here, C_o is annual ordering cost in currency per year; D and Q are in units per year and units per order respectively; and S is currency per order. The expression assumes relatively stable annual demand, a constant order cost and replenishment orders that are not constrained by minimum-order, capacity or supplier-calendar conditions.

In retail, S may vary by channel. A store transfer, an automated vendor-managed replenishment and an emergency courier delivery should not be treated as economically identical. Activity-based costing can improve the estimate by assigning labour and handling resources to the relevant order type.

1.3 Holding cost

Holding cost includes the opportunity cost of capital, warehousing, insurance, shrinkage, handling, deterioration and risk of markdown or obsolescence. If average inventory is I units, unit cost is c, and the annual holding-rate percentage is h, annual holding cost is:

The holding rate h is dimensionless; c is currency per unit; and I is units. If a continuous replenishment system has no safety stock and receipts arrive instantly, average cycle inventory is approximately Q ÷ 2, giving the special-case relationship:

This approximation is not appropriate when demand is intermittent, replenishment is gradual, safety stock is material, inventory is unevenly distributed across locations, or planned promotions create deliberate build-up. A more realistic average inventory may be cycle stock plus safety stock plus pipeline, seasonal and presentation inventory.

The opportunity cost of capital deserves attention. If the retailer’s relevant annual cost of capital is 12%, holding a $100 unit for one year has an approximate capital cost of $12 before storage, insurance and risk costs. The rate should reflect the decision context and should not be added again if it is already embedded in a comprehensive carrying-rate estimate.

1.4 The basic order-quantity trade-off

Under the classical economic order quantity model, annual relevant cost is:

where H is annual holding cost per unit, normally H = c × h. Differentiating with respect to Q and setting the first derivative to zero gives:

The model assumes known and constant demand, constant lead time, instantaneous replenishment, no shortages, no quantity discounts and a constant unit cost. These assumptions are rarely fully true in retail. The formula is therefore a benchmark for understanding the trade-off, not an automatic purchase instruction.

Worked example. Annual demand is 24,000 units, ordering cost is $75 per order and annual holding cost is $3 per unit. Thus:

Q* = √[(2 × 24,000 × 75) ÷ 3] = √1,200,000 ≈ 1,095 units.

Annual ordering cost is (24,000 ÷ 1,095) × 75 ≈ $1,644. Annual cycle-stock holding cost is (1,095 ÷ 2) × 3 ≈ $1,643. The near equality is a property of the unconstrained optimum under the model, not a universal retail rule. A supplier minimum of 1,500 units, limited backroom capacity or a short product life may make the calculated quantity infeasible.

1.5 Shortage and stockout cost

A shortage occurs when available inventory cannot satisfy demand at the required time and location. Its cost may include contribution margin lost, customer switching, substitutions, cancelled orders, expedited replenishment, service recovery and long-term loss of trust. Some effects are observable in transaction data; others are latent and require customer research or controlled experiments.

A useful expected-shortage expression is:

This is a simplified decision expression. In a fuller model, shortage cost is calculated across demand states and shortage quantities. It should not be confused with the accounting value of inventory lost through shrinkage. A stockout concerns insufficient available stock; shrinkage concerns unexplained physical loss.

Shortage costs are asymmetric. A stockout of a destination product, a prescription-related item or a key basket driver may be much more damaging than a stockout of a discretionary accessory. Segmentation should therefore incorporate strategic importance, not merely sales value.

1.6 Markdown and obsolescence cost

Markdown cost is the margin sacrifice associated with selling below the planned price. If planned price is p, markdown price is m, and units sold under markdown are u, the gross-margin sacrifice relative to the planned price is:

This is a managerial measure, not necessarily an accounting loss. It excludes other effects such as incremental demand generated by the lower price and avoided future holding cost. A sound markdown decision compares the value of selling now with the expected value of waiting.

Obsolescence is the loss of economic usefulness caused by fashion change, technology replacement, expiry, regulation, damage or permanently changed demand. It may result in a write-down, disposal cost, donation, recycling cost or liquidation discount. Markdown and obsolescence are related but not identical: markdown is an action or price outcome, whereas obsolescence describes a loss of future usability.

A practical expected-loss formulation is:

The probability should be estimated by product age, lifecycle stage, seasonality, demand volatility and external signals. For regulated or perishable goods, expiry constraints may require first-expiry-first-out controls rather than conventional economic order quantities.

2. Measuring inventory performance

No single metric represents inventory effectiveness. Financial metrics assess capital productivity, operational metrics assess flow and replenishment, and customer metrics assess service. Every metric requires a defined scope, time period, valuation basis and data source.

2.1 Inventory turnover

Inventory turnover indicates how many times average inventory is converted through sales or cost of sales during a period. The preferred financial form is:

Average inventory is commonly calculated as (opening inventory + closing inventory) ÷ 2 when only two observations are available. A monthly average is more reliable for seasonal retail because a year-end balance can be unusually high or low. Turnover is dimensionless, expressed as times per year when the numerator and denominator cover a year.

Using sales value in the numerator and cost-valued inventory in the denominator produces a distorted ratio unless the valuation bases are deliberately reconciled. Turnover can rise because demand improved, inventory fell, or both; interpretation therefore requires sales and service context.

2.2 Days of supply and days inventory outstanding

Days of supply estimates how long current inventory may support expected demand:

If annual demand is used, average daily demand equals annual demand ÷ 365, so:

The numerator and denominator must use compatible units. If inventory is 1,200 units and expected annual demand is 14,600 units, days of supply = 1,200 ÷ (14,600 ÷ 365) = 30 days. This measure is forward-looking when forecast demand is used and historical when past demand is used. It does not account automatically for lead time, seasonality, minimum presentation stock or demand surges.

Days inventory outstanding is a related financial measure, often calculated as average inventory ÷ cost of goods sold × number of days in the period. Days of supply is generally operational and quantity-based; days inventory outstanding is generally financial and value-based.

2.3 Sell-through

Sell-through measures the proportion of available units sold during a specified period:

This form applies when the denominator represents units available for the period after accounting for receipts and adjustments. For a seasonal launch, a retailer selling 700 units with 300 units remaining has sell-through of 700 ÷ 1,000 × 100 = 70%.

Definitions vary. Some organisations use units sold ÷ units received, while others use sales ÷ opening inventory plus receipts. The definition must be documented, especially where returns, transfers, cancellations or stock corrections are material. High sell-through is not automatically good if it was achieved through stockouts or inadequate initial allocation.

2.4 Gross margin return on inventory investment

GMROI assesses gross margin generated for each currency unit invested in average inventory:

Gross margin is sales minus cost of goods sold. If sales are $500,000 and cost of goods sold is $300,000, gross margin is $200,000. If average inventory at cost is $100,000, GMROI is 2.0: each dollar of average inventory generated two dollars of gross margin during the period.

GMROI is especially useful for comparing categories with different prices and margins. It can improve through higher margin, faster turnover or both. However, it may conceal poor customer availability, selling-space constraints, promotional support costs and category roles such as traffic generation.

2.5 Stock-to-sales ratio

The stock-to-sales ratio compares inventory value with sales value for the same period and basis:

If beginning-of-month stock is $180,000 and monthly sales are $300,000, the ratio is 0.60. This ratio is useful for open-to-buy and assortment planning, but it is sensitive to markdowns, seasonality, gross-margin assumptions and whether stock is measured at cost or retail. A falling ratio may indicate improved productivity, but may also signal understocking.

2.6 Fill rate and availability

Fill rate measures the proportion of requested quantity supplied immediately:

If 9,600 of 10,000 requested units are supplied immediately, fill rate is 96%. The unit of analysis must be stated: order lines, units, cases, value or customer orders. A line fill rate can be high while a unit fill rate is low if shortages affect large-quantity orders.

Availability is the proportion of time or demand opportunities in which the product can be purchased in the intended channel and location. A time-based availability measure is:

On-shelf availability is not identical to system inventory. A point-of-sale record may show stock while the item is misplaced, damaged, unprocessed in the backroom or unavailable for online promise. Measurement therefore benefits from combining perpetual records, shelf audits, cycle counts and customer-order outcomes.

3. Interpreting metrics as a balanced system

Metrics can conflict. Consider two categories:

Retail dashboards should combine at least one capital metric, one flow metric, one margin metric and one service metric. For example, turnover with GMROI, days of supply and availability is more informative than turnover alone. Targets should be category-specific and seasonally adjusted; a universal target can encourage harmful behaviour.

4. SKU segmentation and differentiated control

Segmentation converts a large assortment into manageable policy groups. The purpose is not to label products permanently but to assign differentiated rules for forecasting, review frequency, safety stock, audit intensity, allocation and markdown timing.

4.1 ABC analysis: annual usage value

ABC analysis ranks SKUs by annual usage value, commonly:

SKUs are ordered from highest to lowest value and cumulative percentages are calculated. A common, adjustable policy is A: approximately 70–80% of cumulative value; B: the next 15–25%; C: the remaining 5–10%. These percentages are conventions, not laws. Boundaries should reflect managerial capacity and the consequences of error.

A items often receive frequent review, accurate records, tighter approval and close supplier management. C items may use simpler controls, larger review intervals or lower-cost counting methods. ABC is a value prioritisation tool, not a demand-variability or service-criticality tool. A low-value but essential item can be strategically important.

4.2 XYZ analysis: demand predictability

XYZ classifies demand according to variability or forecastability. A common coefficient of variation is:

where σ is the standard deviation of demand observations and μ is their mean over a defined period. CV has no units. Low CV indicates relatively stable demand and high CV indicates greater relative variation. A possible policy is X for low CV, Y for moderate CV and Z for high CV, but thresholds must be calibrated to the category and data frequency.

Intermittent demand requires care. A mean close to zero can make CV unstable, and many zero-demand periods may require measures such as demand occurrence, average inter-demand interval or specialised forecasting methods. XYZ should never be applied without checking whether the data contain stockout-censored demand: observed sales may understate true demand when inventory was unavailable.

4.3 FSN analysis: movement and ageing

FSN classifies products as fast-moving, slow-moving or non-moving using issue frequency, units sold, transaction count or time since last movement. The definition must be explicit. A product sold once in a large quantity is fast by units but not necessarily fast by transaction frequency.

FSN is valuable for identifying dormant stock, location problems and candidates for transfer or markdown. It can misclassify seasonal items as non-moving during their off-season. Ageing should therefore be interpreted against lifecycle and seasonal calendars.

4.4 Multi-criteria segmentation

No single classification captures value, uncertainty, movement and strategic role. A practical multi-criteria model combines dimensions such as:

• financial importance, often ABC;
• demand predictability, often XYZ;
• movement or age, often FSN;
• gross margin or GMROI;
• customer criticality and substitution difficulty;
• lifecycle, perishability and regulatory exposure;
• supplier lead time and replenishment flexibility.

A simple weighted score can support prioritisation:

Here, Score_i is the score for SKU i; w_j is the weight assigned to criterion j; r_{ij} is the normalised rating of SKU i on criterion j; and weights sum to 1. Ratings must have a documented direction: for example, a high shortage consequence should receive a high risk rating, while a high forecastability score may represent either lower or higher control priority depending on the policy. Weighted scores are decision aids, not objective truths. Sensitivity analysis should test whether small changes in weights alter the resulting segment.

A useful policy matrix might distinguish A-X items, which are high-value and predictable, from A-Z items, which are high-value and volatile. A-X products may support automated replenishment with tight records; A-Z products may require senior review, shorter review cycles and scenario-based allocation. C-Z products may warrant minimum display quantities, supplier-to-order arrangements or rationalisation, unless they are strategically critical.

Figure: Illustrative policy matrix combining value and demand uncertainty.

5. Applied case: Northstar Department Stores

The following is an illustrative case. Northstar operates 18 department stores and an online channel. Its winter fashion category shows 68% sell-through after a 35% average markdown. Inventory turnover improved from 3.4 to 4.1 times, but online availability fell to 91% for several high-traffic sizes. Managers celebrated the turnover improvement while finance reported declining category GMROI.

Background and problem

The category buyer had used a single six-week replenishment rule. ABC analysis identified premium coats as A items, but no XYZ or lifecycle classification was used. Forecast demand was based on sales history without adjustment for stockouts. Store transfers were recorded with delays, so the system showed stock in stores where customers could not purchase it.

Analysis

A review separated planned-price margin from markdown margin and measured inventory at cost. High turnover was partly caused by selling fewer units at lower prices and closing the season with low stock. The category’s 91% online availability concealed size-level availability below 85%. Premium coats were A-Z items: financially important but volatile and fashion-sensitive. Basic thermal garments were B-X items: predictable, replenishable and suitable for automated reorder points. Several accessories were C-FSN non-moving items despite acceptable historical sales because the season had ended.

Approach and solution

Northstar introduced three controls. First, it measured GMROI, availability and sell-through together, using weekly seasonal cohorts. Second, it applied differentiated segmentation: A-Z items received smaller initial commitments, more frequent review and earlier option-based supplier replenishment; B-X items used forecast-driven replenishment; C non-moving items entered a transfer, bundle or markdown workflow. Third, the retailer introduced store-level cycle counts for online-eligible inventory and reconciled transfer confirmations within 24 hours.

Markdown decisions were moved forward for products whose projected end-of-season stock exceeded planned demand, but price reductions were restrained for high-availability basics. The company also recorded lost-sales signals and distinguished a genuine demand decline from a stockout-censored sales history.

Outcome

After one season, the category achieved 95% online availability, reduced late-season markdown depth to 24%, and raised GMROI despite a modest reduction in unit sell-through. The result illustrates that a lower sell-through percentage can coexist with better economic performance when full-price sales, availability and inventory timing improve.

Learning points

1. Turnover should not be rewarded without considering margin and service.
2. ABC alone cannot identify volatility or strategic criticality.
3. System inventory is not equivalent to purchasable inventory.
4. Lifecycle-aware segmentation improves markdown timing.
5. Data quality, especially transfer and stockout data, is part of inventory performance.

6. Governance, technology and contemporary practice

Inventory metrics require governance. Organisations should document definitions, valuation methods, calendars, exclusions, ownership and correction procedures. A fill-rate dashboard without a stated unit basis invites misleading comparisons. Similarly, a GMROI measure must specify whether average inventory is at cost, retail or a blended valuation.

Technology can improve execution through RFID, computer vision, electronic shelf labels, demand sensing, digital twins and machine-learning forecasts. These tools do not eliminate judgement. Algorithms trained on stockout-affected sales may learn under-ordering; automated markdowns may disadvantage customers who cannot purchase at the right time; and opaque supplier or platform data can weaken auditability. Human review is especially important for A-Z items, regulated products, ethical sourcing concerns and unusual demand events.

Contemporary retailers increasingly use probabilistic forecasts, scenario planning and omnichannel inventory pools. Rather than relying on one forecast, managers examine a distribution of possible demand and the cost of being over or under. Inventory may also be positioned for total network contribution rather than store-level efficiency. A unit held centrally can serve several locations, but it may increase delivery time and fulfilment cost. The correct decision is therefore network-specific.

Best practice includes:

• define every metric and preserve unit consistency;
• use cost and service metrics together;
• segment by value, variability, movement and strategic role;
• adjust thresholds for seasonality and category economics;
• distinguish observed sales from unconstrained demand;
• audit physical, system and online-available inventory;
• test classification sensitivity and review segment membership periodically;
• assign decision rights for replenishment, markdown and exception approval;
• monitor unintended incentives, such as suppressing orders to improve turnover;
• retain an auditable record of assumptions, overrides and data corrections.

Chapter Summary

Inventory economics concerns the balance among ordering, holding, shortage, markdown and obsolescence costs. The classical order-quantity model clarifies the ordering–holding trade-off, but its assumptions must be tested against retail realities such as seasonality, uncertain demand, minimum orders and product lifecycles.

Inventory performance requires a portfolio of measures. Turnover and days of supply describe capital and flow; sell-through describes the conversion of available units; GMROI links gross margin with inventory investment; stock-to-sales supports planning; and fill rate and availability assess service. Definitions, valuation bases and time periods must be explicit, and high performance on one metric may conceal failure on another.

ABC identifies financial importance, XYZ identifies demand predictability, and FSN identifies movement or inactivity. Multi-criteria segmentation integrates these dimensions with margin, strategic criticality, lifecycle, lead time and substitution risk. Segments should lead to differentiated policies for review, replenishment, audit, allocation and markdown. Technology can strengthen visibility and forecasting, but governance, data quality and managerial judgement remain essential.

Key Terms

Inventory economics: Analysis of the costs and benefits of holding and replenishing stock.

Ordering cost: Cost incurred each time an order or replenishment event is initiated and processed.

Holding cost: Cost of financing, storing, handling, insuring and risking inventory over time.

Shortage cost: Economic and service loss caused by insufficient inventory when demand occurs.

Markdown cost: Margin sacrifice associated with selling merchandise below its planned price.

Obsolescence: Loss of economic usefulness caused by age, expiry, fashion, technology or changed demand.

Inventory turnover: Cost of goods sold divided by average inventory at cost.

Days of supply: Inventory on hand divided by expected average daily demand.

Sell-through: Units sold as a proportion of units available for sale.

GMROI: Gross margin generated per currency unit invested in average inventory.

Stock-to-sales ratio: Inventory value divided by sales value for a defined period and basis.

Fill rate: Proportion of requested quantity supplied immediately.

Availability: Proportion of time or demand opportunities in which an item is purchasable.

ABC analysis: Classification by annual usage value or another financial-value measure.

XYZ analysis: Classification by demand variability or predictability.

FSN analysis: Classification by movement speed or inventory age.

Coefficient of variation: Standard deviation divided by the mean.

SKU segmentation: Grouping products to assign differentiated management policies.

Stockout-censored demand: Observed sales that understate true demand because inventory was unavailable.

Review Questions

Short Answer Questions

1. Why can a high inventory turnover ratio be accompanied by poor retail performance?
2. Distinguish markdown cost from obsolescence cost.
3. What does GMROI measure, and why is inventory valued at cost in its denominator?
4. State one limitation of ABC analysis.
5. Why must fill rate specify its unit of analysis?

Long Answer Questions

1. Explain the ordering–holding cost trade-off and derive the classical economic order quantity under its assumptions.
2. Compare turnover, days of supply, sell-through and stock-to-sales ratio, including the managerial question addressed by each.
3. Discuss how shortage, markdown and obsolescence costs should influence seasonal retail decisions.
4. Explain ABC, XYZ and FSN analysis and propose differentiated controls for each classification.
5. Evaluate the benefits and risks of using AI-enabled forecasting and automated markdown systems in inventory management.

Analytical / Application Questions

1. A category has annual demand of 18,000 units, ordering cost of $60 per order and annual holding cost of $2.40 per unit. Calculate the classical economic order quantity and explain two reasons why the result might not be implemented exactly.
2. A retailer sells 7,500 units during a launch and has 2,500 units remaining. Calculate sell-through and explain why the result alone cannot establish that the launch was successful.
3. Design a multi-criteria segmentation policy for two products: a high-margin, volatile fashion item and a low-margin, predictable staple. Specify the metrics, review frequency, replenishment logic and service priorities you would use.

Multiple Choice Questions with Answers

Question 1

Which expression represents annual inventory turnover using a consistent financial valuation basis?

A. Average inventory at cost ÷ cost of goods sold

B. Cost of goods sold ÷ average inventory at cost

C. Sales value ÷ ending units on hand

D. Units sold ÷ ordering cost

Correct Answer: B. Cost of goods sold ÷ average inventory at cost

Question 2

Annual demand is 14,600 units and inventory on hand is 1,200 units. Using 365 days, what is the days of supply?

A. 12 days

B. 24 days

C. 30 days

D. 44 days

Correct Answer: C. 30 days

Question 3

A retailer sells 700 units and has 300 units remaining from the same available pool. What is sell-through?

A. 30%

B. 57%

C. 70%

D. 100%

Correct Answer: C. 70%

Question 4

Sales are $500,000, cost of goods sold is $300,000, and average inventory at cost is $100,000. What is GMROI?

A. 1.5

B. 2.0

C. 3.0

D. 5.0

Correct Answer: B. 2.0

Question 5

Which classification primarily identifies demand predictability or variability?

A. ABC

B. XYZ

C. FSN

D. GMROI

Correct Answer: B. XYZ

Question 6

In the classical economic order quantity model, which change generally increases the calculated order quantity, all else equal?

A. Higher annual holding cost per unit

B. Lower annual demand

C. Higher ordering cost

D. Lower unit demand variability

Correct Answer: C. Higher ordering cost

Question 7

A fill rate of 96% means that:

A. 96% of inventory value earned gross margin

B. 96% of requested quantity was supplied immediately

C. 96% of time the store had no inventory

D. 96% of items were sold at full price

Correct Answer: B. 96% of requested quantity was supplied immediately

Question 8

Why can a low-value SKU still require high inventory control priority?

A. ABC always classifies it as an A item

B. Its strategic criticality or substitution difficulty may be high

C. Its coefficient of variation must be zero

D. Low value automatically increases obsolescence

Correct Answer: B. Its strategic criticality or substitution difficulty may be high

Question 9

If gross margin is $200,000 and average inventory at cost is $100,000, GMROI is:

A. 0.5

B. 1.0

C. 2.0

D. 3.0

Correct Answer: C. 2.0

Question 10

Which statement best describes stockout-censored demand?

A. Demand recorded only after a markdown

B. Observed sales that understate potential demand because stock was unavailable

C. Demand caused by obsolete products

D. Demand measured only in currency values

Correct Answer: B. Observed sales that understate potential demand because stock was unavailable