Merchandising, Assortment, Pricing and Promotion Systems

Introduction

Retail technology has changed merchandising from a periodic, spreadsheet-based activity into an integrated decision system. Contemporary retailers combine point-of-sale data, inventory records, customer information, supplier data, market intelligence and analytical models to decide what to sell, where to place it, how much to buy, when to replenish it, what price to charge and which promotion to offer.

The central managerial problem is not simply sales maximisation. A successful retail system balances revenue, gross margin, stock availability, working capital, customer value and operational feasibility. An assortment that produces high sales but excessive markdowns may destroy profitability; a highly profitable assortment that is frequently out of stock may damage customer loyalty. Retail technology therefore supports coordinated decisions across merchandising, planning, buying, allocation, pricing and promotion.

This chapter examines the architecture and application of these systems, with particular attention to merchandise planning, category management, assortment optimisation, allocation, markdown management, dynamic pricing and promotion analytics.

Learning Objectives

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

1. Explain the role and architecture of merchandising and pricing systems.
2. Apply merchandise-planning and category-management concepts to retail decisions.
3. Analyse assortment breadth, depth, localisation and SKU productivity.
4. Calculate fundamental retail performance, inventory and markdown measures.
5. Evaluate allocation, replenishment, dynamic pricing and promotion decisions.
6. Assess the managerial, ethical, technological and governance implications of retail analytics.

1. The Retail Decision System

A retail decision system links strategic intent with operational execution. Its principal layers are:

The system depends on data quality. Product identifiers, store hierarchies, size and colour attributes, cost prices, regular prices, promotion flags, stock-on-hand and sales timestamps must be consistent. A technically sophisticated model using inaccurate inventory or incorrectly classified promotions can produce systematically poor decisions.

A useful architecture separates the system of record from the system of decision. The former stores transactions, products, suppliers and inventory. The latter transforms these data into forecasts, recommendations, constraints and scenarios. The final decision may still require merchant judgement because strategic brand considerations, supplier relationships and emerging trends are not completely observable in historical data.

2. Merchandise Planning and Financial Control

Merchandise planning translates business strategy into financial and inventory targets by department, category, channel, location and time period. A plan normally includes sales, cost of goods, gross margin, receipts, inventory, markdowns and stock turn.

The basic retail profit identity is:

Net sales are sales after returns, discounts and allowances. Gross margin percentage is:

The denominator is net sales; therefore gross margin percentage should not be confused with markup percentage, which is commonly calculated relative to cost:

For example, a product costing $60 and selling for $100 has a $40 gross profit, a 40% gross margin and a 66.7% markup. These measures answer different questions and must not be substituted for one another.

A merchandise financial plan begins with a sales and margin target and translates it into an inventory investment. The stock-to-sales ratio for a period is:

The inventory turnover measure is:

Average inventory is often approximated as opening inventory plus closing inventory divided by two. The approximation is less reliable for highly seasonal businesses, where monthly or weekly averages are preferable. Turnover is meaningful only when inventory and cost of goods sold are measured on the same cost basis and over the same period.

A complementary measure is gross margin return on inventory investment (GMROII):

If a category generates $240,000 gross margin and requires average inventory of $120,000 at cost, its GMROII is 2.0. This means that each dollar of average inventory investment generated two dollars of gross margin during the period. GMROII is useful for comparing categories, but comparisons should account for seasonality, lead time, shrinkage, space and strategic role.

Open-to-Buy Control

Open-to-buy (OTB) is the purchasing capacity remaining after considering the plan, existing inventory and committed receipts. A simplified retail-value version is:

Here all terms must be expressed at the same retail valuation and for the same planning period. The formula is a control approximation; returns, transfers, cancellations, shrinkage and valuation conventions may require additional adjustments. A positive OTB indicates capacity for further buying, not an automatic instruction to buy.

3. Category Management

Category management treats a group of products as a strategic business unit because shoppers often perceive categories rather than individual SKUs. A category may be assigned a role such as:

• Destination: attracts shoppers and establishes retailer authority.
• Routine: satisfies frequent, predictable needs.
• Seasonal or occasion: supports periodic demand.
• Convenience: enables easy completion of a shopping mission.
• Image or excitement: creates differentiation and discovery.

The role influences assortment, price architecture, space, service level and promotional intensity. Category management is not identical to supplier management. A supplier may provide information and funding, but the retailer remains responsible for customer value, competitive neutrality and total category performance.

A category review commonly follows this sequence:

1. Define the category and its shopper mission.
2. Assess internal performance and external market conditions.
3. Assign a role and measurable objectives.
4. Develop assortment, pricing, shelving, supply and promotion tactics.
5. Implement through stores and digital channels.
6. Measure results and revise the plan.

Important diagnostic measures include sales per SKU, gross margin per SKU, sell-through, availability, weeks of supply, return rate, markdown rate, basket attachment and customer conversion. A category dashboard should distinguish demand failure from execution failure. Low sales can result from weak product appeal, poor placement, inaccurate stock records or out-of-stock conditions.

4. Assortment Planning and Optimisation

Assortment planning determines the products offered in a particular channel, store cluster, website or customer segment. Four dimensions are especially important:

• Breadth: number of categories or product families.
• Depth: number of options within a category, such as brands, sizes, colours or price points.
• Localisation: adaptation to geography, climate, demographics, culture and store mission.
• Channel coherence: the relationship between physical, online, marketplace and fulfilment assortments.

More SKUs can increase choice and reduce substitution losses, but they also increase inventory exposure, operational complexity and cognitive burden. The objective is not maximum variety; it is the best customer and financial outcome under constraints.

An assortment model may use a binary decision variable x_i ∈ {0,1}, where x_i = 1 means SKU i is included. A simplified contribution-maximisation formulation is:

subject to space, inventory, budget, minimum presentation and category constraints. Here mᵢ is expected contribution margin from SKU i, hᵢ is the relevant holding or operating cost, and n is the number of candidate SKUs. This is a simplified model: realistic systems also include substitution, cannibalisation, size curves, supplier minimums, demand uncertainty and strategic requirements.

A product’s incremental value should not be judged solely by its own sales. If a new premium SKU attracts customers and increases basket value, its total contribution may exceed its direct margin. Conversely, two products with similar demand may cannibalise one another. Assortment optimisation therefore benefits from choice modelling, demand elasticity estimates and controlled tests rather than simple ranking by historical sales.

Sell-Through and Weeks of Supply

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

The formula assumes that the numerator and denominator use the same product scope and period. If transfers or stock adjustments are material, they must be incorporated into the available-unit definition.

Weeks of supply is an estimate of how long current stock will last:

It assumes that the forecast is positive and that no major receipt, seasonality shift or capacity constraint changes the demand rate. A high weeks-of-supply value may indicate overstock; a low value may indicate an impending stockout, although the appropriate threshold differs by lead time and category.

5. Forecasting, Allocation and Replenishment

Forecasting systems estimate demand at product, location, channel and time-period levels. Methods may include moving averages, exponential smoothing, regression, hierarchical forecasting and machine learning. No algorithm eliminates uncertainty. Forecasts should be evaluated using measures such as mean absolute error (MAE):

where Aₜ is actual demand, Fₜ is forecast demand and n is the number of observations. MAE is expressed in units when demand is measured in units. Percentage measures can be misleading when actual demand is zero or very small.

Allocation decides how a finite quantity is distributed among stores or fulfilment locations. A basic allocation may use historical sales, forecast demand, store capacity, presentation minimums and local attributes. A more advanced model minimises expected lost margin and holding cost while meeting service constraints.

Allocation should distinguish initial allocation from replenishment. Initial allocation supports launch and presentation; replenishment responds to realised sales and updated forecasts. A store with historically low sales may deserve more stock if its low sales were caused by chronic stockouts. This is a classic endogeneity problem: observed sales may underestimate unconstrained demand.

A service-level decision can incorporate safety stock. Under a simple independent-demand approximation:

If demand during lead time is normally distributed and a target cycle service level is selected, safety stock is often represented as zσ_LT, where z is the standard-normal quantile and σ_LT is the standard deviation of demand during lead time. This representation assumes an appropriate distributional approximation and a correctly estimated lead-time demand variance. It should not be applied mechanically when demand is intermittent, highly seasonal or constrained by availability.

6. Pricing Systems

Pricing systems combine cost, customer willingness to pay, competition, inventory position, product life cycle, channel policy and strategic objectives. A price decision can be described through a price ladder: opening price point, good-better-best tiers, premium anchors and promotional prices.

Price elasticity of demand measures the percentage response in quantity to a percentage change in price:

Using a midpoint or arc formulation for two observations reduces dependence on the direction of comparison:

Here P₁, P₂ are prices and Q₁, Q₂ are corresponding quantities. Demand is commonly described as elastic when the absolute value of elasticity exceeds one and inelastic when it is below one. However, historical price changes may be confounded by promotions, seasonality, competition and stock availability; observed correlation is not automatically causal elasticity.

If unit variable cost is c and price is p, unit contribution is p − c. A price increase can raise contribution per unit but reduce volume. The relevant decision compares total contribution, inventory effects, customer response and long-term reference-price effects.

Markdown Management

Markdowns are deliberate price reductions used to clear inventory, stimulate demand or manage product life-cycle risk. A markdown system should identify the appropriate timing, depth and scope rather than simply reducing price when sales are weak.

Markdown rate can be expressed as:

A markdown decision should consider remaining stock, forecast demand, time to season end, replenishment status, gross margin, salvage value and the opportunity cost of occupying space. Early markdowns may improve sell-through but sacrifice margin on units that would have sold at the original price. Late markdowns preserve initial margin but increase residual inventory risk.

Markdown optimisation is especially difficult when products are perishable, fashion-sensitive or size-specific. A system should model the probability of selling at alternative prices and allow merchant override with an audit trail.

Dynamic Pricing

Dynamic pricing changes prices in response to relevant conditions, such as demand, inventory, time, competitor prices, capacity or customer segment. It is suitable where demand varies rapidly and price changes can be communicated and governed effectively. It is less suitable where price consistency, trust, regulation or operational stability is strategically important.

A responsible dynamic-pricing system requires:

• clearly defined objectives and constraints;
• reliable, timely data;
• price floors, ceilings and change-frequency limits;
• monitoring for discriminatory or geographically unfair outcomes;
• separation of legitimate demand signals from protected characteristics;
• human oversight and an explanation or appeal process where appropriate;
• experiment design that identifies causal effects.

Frequent price changes can create consumer confusion and perceived unfairness. Competitor scraping can also create legal and operational risks if data are inaccurate or collected inappropriately. Dynamic pricing should therefore be treated as a governed decision process, not an autonomous revenue-maximisation mechanism.

7. Promotion Management and Evaluation

Promotions include temporary price reductions, multibuy offers, coupons, loyalty rewards, displays, bundles and personalised incentives. Their economic value depends on incremental demand, margin dilution, operational cost and post-promotion effects.

A promotion should be evaluated against a credible baseline. Incremental units are:

A simplified incremental gross profit is:

Promotion costs may include media, funding, display, fulfilment, coupon redemption and additional labour. Supplier funding should be recorded transparently rather than treated as costless revenue.

Suppose a product normally sells 1,000 units at $10, with unit cost $6. During a promotion it sells 1,400 units at $8. Promotion execution costs are $400. Baseline gross profit is $4,000. Promoted gross profit is (8 − 6) × 1,400 = 2,800. After execution cost it is $2,400. Incremental units are 400, and incremental gross profit relative to the baseline is therefore −$1,600. The promotion increased unit sales but reduced gross profit. This conclusion may change if the product drives profitable basket attachment or customer acquisition, which must be measured separately.

Promotion effectiveness requires causal measurement. Suitable designs include holdout stores, matched markets, customer-level randomisation and difference-in-differences analyses. A simple before-and-after comparison is weak because it cannot separate promotion effects from seasonality, trends, competitor actions or changes in availability.

Common promotion problems include forward buying, pantry loading, cannibalisation, promotion fatigue, training customers to wait for discounts and inaccurate baseline forecasts. Promotion-management systems should integrate offer setup, eligibility, funding, legal terms, POS execution and post-event analysis.

8. Technology, Data and AI

Modern retail platforms increasingly use machine learning for demand forecasting, recommendation, price optimisation, image recognition, assortment localisation and anomaly detection. Generative AI may assist merchants by summarising category performance, drafting scenario explanations or querying data using natural language. It should not be allowed to create unverified prices, product claims or customer segments without controls.

A sound analytical operating model includes:

1. Data governance: ownership, definitions, quality rules and lineage.
2. Model governance: validation, performance monitoring, drift detection and version control.
3. Decision governance: approval rights, override rules and auditability.
4. Experiment governance: pre-specified hypotheses, treatment definitions and success metrics.
5. Responsible-use governance: privacy, fairness, security, explainability and regulatory compliance.

Personalised promotions require a lawful basis for data use, purpose limitation, appropriate consent or notice, security controls and retention policies. Sensitive attributes and proxy variables require particular caution. A model that gives systematically worse offers or availability to a protected group may create discrimination even if the protected attribute is not explicitly included.

9. Integrated Applied Case: Northstar Apparel

The following is an illustrative case. Northstar Apparel, a mid-sized omnichannel retailer, carries 2,000 fashion SKUs across stores and online. It has strong full-price sales but frequent end-of-season markdowns. Store managers complain that allocations do not reflect local climate and that popular sizes sell out early.

Background and Problem

The business used a national assortment, monthly spreadsheets and merchant-led markdowns. Inventory data were delayed, online and store stock records were not consistently reconciled, and promotion results were reported as sales uplift without a baseline. Management sought higher gross margin and better availability without increasing total inventory.

Analysis

The analytical team found that:

• 18% of SKUs generated less than 2% of sales but consumed substantial presentation space;
• sell-through differed markedly by climate zone;
• several apparent low-demand stores had persistent stockouts in key sizes;
• promotions increased units but often reduced category gross profit;
• markdown timing varied by manager and had no documented decision rule.

The team segmented stores by climate, size profile, traffic and digital fulfilment capability. It created a common product hierarchy and reconciled stock-on-hand with cycle counts. Rather than ranking products only by sales, it estimated contribution, substitution and local demand.

Approach and Solution

Northstar introduced a six-part system:

1. A category-level financial plan with sales, margin, inventory and markdown targets.
2. Localised core assortments plus a wider online long tail.
3. Size- and store-specific initial allocation using forecast demand and presentation minimums.
4. Weekly replenishment based on unconstrained demand indicators and available inventory.
5. A markdown calendar using weeks remaining, stock cover and forecast sell-through.
6. Promotion tests using matched stores and incremental gross profit as the primary measure.

The system imposed price floors, required merchant approval for exceptional markdowns and retained an audit record of model recommendations and overrides. A dashboard separated genuine demand decline from stockout-related lost sales.

Outcome

After two seasons, management reported improved availability in core sizes, fewer emergency transfers and a lower residual stock position. Full financial effects were assessed against a matched control group rather than a simple year-on-year comparison. The strongest improvement came from better allocation and stock accuracy, not from more aggressive discounting.

Learning Points

The case demonstrates that pricing and promotion cannot compensate for poor assortment and allocation. It also shows why integrated data architecture is essential: a markdown model is unreliable when stock records are wrong, and an assortment model is unreliable when sales are censored by stockouts. Finally, technology improved decisions only after merchants, analysts and operations teams agreed on definitions, authority and performance measures.

10. Challenges and Limitations

Retail systems face several persistent limitations:

• Data sparsity: new products and low-volume stores have little history.
• Censored demand: sales do not reveal demand when products are unavailable.
• Substitution and cannibalisation: SKU-level forecasts may double-count demand.
• Concept drift: customer preferences, competition and economic conditions change.
• Organisational conflict: buying, stores, finance and suppliers may have different incentives.
• Model opacity: complex systems can make recommendations that merchants cannot challenge effectively.
• Execution gaps: a correct recommendation fails if labels, shelf space, inventory or POS configuration are wrong.
• Strategic externalities: short-term margin optimisation may harm brand equity or customer trust.

Managers should use scenario analysis and sensitivity testing. For example, the effect of a price change should be assessed under alternative elasticity estimates rather than treated as a single certain prediction. Human judgement is most valuable when evidence is weak, the situation is novel or the decision has substantial reputational consequences.

Best-Practice Framework

A mature merchandising and pricing capability should:

1. Establish one product, location, calendar and price vocabulary.
2. Reconcile inventory and distinguish observed sales from unconstrained demand.
3. Set targets at category and channel level before optimising individual SKUs.
4. Use contribution, customer value and inventory productivity together.
5. Localise assortments where evidence supports local differences.
6. Measure promotions against causal baselines and incremental profit.
7. Place explicit constraints on dynamic pricing and automated markdowns.
8. Combine algorithmic recommendations with accountable merchant review.
9. Monitor model accuracy, bias, drift and override patterns.
10. Close the loop by feeding outcomes back into forecasts and plans.

Chapter Summary

Merchandising technology integrates planning, category management, assortment, allocation, pricing, markdown and promotion decisions. Merchandise planning converts strategic objectives into sales, margin and inventory targets. Category management organises products around shopper missions and business roles. Assortment planning balances breadth, depth, localisation and inventory risk, while allocation converts forecasts into location-specific stock decisions.

Retail metrics such as gross margin, turnover, GMROII, sell-through, weeks of supply and OTB support disciplined decisions, provided their definitions, units and assumptions are consistent. Pricing systems must distinguish margin, markup and elasticity, and markdowns should balance full-price opportunity against residual inventory risk. Promotion success depends on incremental, not merely observed, sales and should be evaluated using credible baselines.

AI and dynamic pricing can improve responsiveness, but only within strong data, model, ethical and managerial governance. The most effective retail systems combine analytics with merchant expertise, operational execution and continuous measurement.

Key Terms

Merchandise planning: Planning sales, margin, inventory and receipts by category, channel or period.

Category management: Managing a product category as a strategic business unit based on shopper needs.

Assortment: The set of products offered through a store, channel or location.

Breadth: The number of product categories carried.

Depth: The number of alternatives within a category.

Localisation: Adapting an assortment or decision to local market conditions.

Open-to-buy: Remaining purchasing capacity after considering planned inventory, sales and commitments.

Gross margin: Net sales minus cost of goods sold.

Markup: Selling-price profit expressed relative to product cost.

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

GMROII: Gross margin dollars generated per dollar of average inventory investment.

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

Weeks of supply: Available stock divided by forecast weekly demand.

Allocation: Distribution of inventory among stores or fulfilment locations.

Markdown: A planned reduction from an original selling price.

Dynamic pricing: Price adjustment in response to changing market or operational conditions.

Price elasticity: Percentage change in quantity divided by percentage change in price.

Promotion incrementality: Sales or profit caused by a promotion beyond a credible baseline.

Cannibalisation: Reduction in one product’s demand caused by another product or offer.

Model governance: Controls for the validation, monitoring, use and accountability of analytical models.

Review Questions

Short Answer Questions

1. Distinguish gross margin percentage from markup percentage.
2. What is the purpose of an open-to-buy calculation?
3. Explain the difference between initial allocation and replenishment.
4. Why is a promotion’s sales uplift not sufficient evidence of success?
5. State two governance requirements for dynamic pricing.

Long Answer Questions

1. Explain how merchandise planning, category management and assortment planning are connected.
2. Discuss the advantages and limitations of using machine learning in retail demand forecasting.
3. Explain how a retailer can use GMROII, sell-through and weeks of supply together.
4. Analyse the strategic and operational risks of excessive markdown dependence.
5. Describe a responsible framework for implementing personalised promotions.

Analytical / Application Questions

1. A category has net sales of $500,000, cost of goods sold of $320,000 and average inventory at cost of $160,000. Calculate gross margin percentage, inventory turnover and GMROII, and interpret the results.
2. A promotion sells 1,800 units at $7. The unit cost is $4, normal baseline demand is 1,200 units, and promotion execution costs are $500. Calculate incremental units and incremental gross profit relative to normal sales at the regular price of $8. State any assumptions.
3. A retailer is considering a new SKU that has attractive direct margin but is highly substitutable with an existing SKU. Design an assortment test that measures incremental category contribution rather than only the new SKU’s sales.

Multiple Choice Questions with Answers

Question 1

A product costs $60 and sells for $100. What is its gross margin percentage?

A. 40%

B. 60%

C. 66.7%

D. 166.7%

Correct Answer: A. 40%

Question 2

Which measure divides gross margin dollars by average inventory cost?

A. Sell-through

B. GMROII

C. Markup percentage

D. Stock-to-sales ratio

Correct Answer: B. GMROII

Question 3

What is the principal purpose of allocation?

A. To determine corporate tax expense

B. To distribute available inventory among locations

C. To calculate supplier lead time

D. To replace category roles

Correct Answer: B. To distribute available inventory among locations

Question 4

A promotion sells 1,400 units while baseline demand is 1,000 units. What are the incremental units?

A. 400

B. 1,000

C. 1,400

D. 2,400

Correct Answer: A. 400

Question 5

Which condition most directly weakens a simple before-and-after promotion comparison?

A. A fixed product cost

B. A stable product hierarchy

C. Seasonality occurring during the promotion

D. A documented promotion end date

Correct Answer: C. Seasonality occurring during the promotion

Question 6

Weeks of supply is calculated as available units divided by what quantity?

A. Forecast weekly unit demand

B. Annual gross margin

C. Original retail value

D. Average selling price

Correct Answer: A. Forecast weekly unit demand

Question 7

Which is an example of assortment depth?

A. Adding a new product category

B. Offering more sizes within a footwear category

C. Opening a store in another region

D. Changing the financial year

Correct Answer: B. Offering more sizes within a footwear category

Question 8

What does price elasticity relate?

A. Inventory value to store space

B. Percentage quantity response to percentage price change

C. Gross profit to supplier funding

D. Receipts to lead time

Correct Answer: B. Percentage quantity response to percentage price change

Question 9

Why can observed sales underestimate true demand?

A. Because all sales are always forecast values

B. Because stockouts censor potential sales

C. Because gross margin is always negative

D. Because assortment breadth is zero

Correct Answer: B. Because stockouts censor potential sales

Question 10

Which control is most appropriate for a dynamic-pricing system?

A. Unlimited price changes without review

B. Removal of all price floors

C. Price boundaries and monitoring for unfair outcomes

D. Using protected characteristics directly

Correct Answer: C. Price boundaries and monitoring for unfair outcomes