Markdown, Clearance and Revenue Management

Introduction and Context

Retail pricing does not end when a product is launched. Demand changes as seasons progress, fashion relevance declines, perishability reduces usable life, and selling opportunities disappear when a capacity-constrained service date passes. Retailers must therefore decide not only what price to charge, but also when to reduce price, by how much, and through which channel.

A markdown is a planned reduction from an initial or current selling price. Clearance is the terminal phase in which inventory is liquidated, transferred, returned, donated, recycled, or written off. Revenue management is the systematic allocation of limited, perishable capacity among customers and time periods to maximise expected revenue or contribution. These ideas are closely related, but they are not identical. A fashion retailer manages physical inventory and style risk; an airline manages seats whose value disappears at departure; a supermarket manages products whose quality and safety decline with time.

The managerial challenge is to balance four competing objectives:

1. recover margin and avoid unnecessary price reductions;
2. achieve an acceptable sell-through rate;
3. control inventory ageing, stockholding cost and obsolescence;
4. protect customer trust, brand equity and future reference prices.

Markdown decisions are consequently forecasting, optimisation and governance problems rather than simple promotional decisions.

Learning Objectives

After completing this chapter, the reader should be able to:

• distinguish markdown, clearance, liquidation and revenue management;
• explain the economic logic of markdown timing and depth;
• calculate sell-through, gross margin, stock cover, inventory ageing and markdown cost;
• use demand forecasts, price elasticity and scenario analysis in markdown planning;
• apply revenue-management principles to seasonal, fashion, perishable and capacity-constrained retail;
• evaluate optimisation models, practical constraints, ethics and implementation risks;
• design a controlled markdown process using data, technology and managerial judgement.

1. Core Concepts and Terminology

1.1 Markdown and clearance

A markdown is a price reduction intended to stimulate demand, accelerate stock movement or respond to changed market conditions. It may be temporary or permanent. A clearance price normally signals that the retailer is seeking final disposal rather than maintaining a long-term price position.

Important distinctions are as follows:

The initial price is the launch or ticket price. The current price is the price before a proposed reduction. The markdown depth is commonly expressed as:

Here, P₀ is the reference price and P₁ is the new selling price. Both prices must use the same tax basis and currency. If P₀ = $100 and P₁ = $75, the depth is 25%.

1.2 Sell-through and inventory ageing

Sell-through measures the proportion of available units sold during a defined period. Two conventions are widely used, so managers must state which one is being applied:

“Units available” may mean opening inventory plus receipts, or opening inventory only. The first convention is more suitable for an in-season flow; the second is useful for evaluating a buy. Returns, transfers and cancellations should be treated consistently.

Inventory ageing is the time inventory has been held or the time since receipt. Age bands—for example, 0–30, 31–60 and over 60 days—allow managers to identify increasing obsolescence risk. Stock cover estimates how long current inventory will last:

This calculation assumes that forecast demand is positive and reasonably comparable with the inventory definition. It is not reliable when demand is highly intermittent or the product is unavailable in part of the period.

1.3 Margin, contribution and recovery

Let P be selling price per unit, C_v variable cost per unit, Q units sold and F fixed costs allocated to the decision. Contribution is:

For markdown decisions, avoid treating sunk purchase cost as avoidable. The relevant comparison is usually between the expected incremental contribution from selling now and the expected value of waiting, transferring, returning, donating or writing off the unit. Nevertheless, purchase cost remains important for accounting gross margin and profitability reporting.

The recovery rate is the cash or value recovered from an item relative to a chosen cost basis. The basis—landed cost, book value or expected full-price revenue—must be specified. A clearance can produce a low gross margin but still be economically rational if the alternative is a costly write-off.

2. The Economics of Markdown Timing and Depth

A markdown has two effects. It reduces unit margin, but it may increase the probability and speed of sale. It can also reduce future holding costs and prevent terminal obsolescence. The correct decision compares expected total value, not merely the percentage discount.

Suppose a retailer is considering price P₁ now or waiting one period. A simplified expected-value comparison is:

P₁ and P₂ are prices now and later; q₁ and q₂ are expected units sold; C_v is avoidable variable cost; H and H₂ are relevant holding and operating costs; L is expected loss from obsolescence, expiry, cancellation or residual disposal. The option with the greater expected value is preferred, subject to service, brand and capacity constraints.

A more complete multi-period formulation uses scenarios s with probabilities π_s:

The probabilities must sum to one and scenarios should represent materially different demand, competitor and operational outcomes. The model is only as credible as its forecasts and cost definitions.

2.1 Demand response and elasticity

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

For ordinary goods, ε is often negative. A price cut increases revenue from price response alone only when demand is sufficiently elastic over the relevant range. However, markdown decisions also depend on holding cost, terminal value, inventory availability and customer segmentation. Elasticity estimated from a broad historical average can mislead when discounts alter product visibility, urgency or customer composition.

For a linear local approximation, if demand is Q = a − bP, revenue is R = P(a − bP). The unconstrained revenue-maximising price is:

This is a special case, assuming linear demand, constant parameters and no inventory, capacity or competitive reaction. It should not be used mechanically for seasonal clearance.

2.2 Timing logic

A markdown should generally be considered earlier when:

• demand uncertainty is high and the product has a short selling window;
• inventory is above the sales plan and replenishment is difficult to stop;
• the cost of terminal disposal or transfer is high;
• competitors have already reset reference prices;
• fashion relevance or quality declines rapidly.

It may be delayed when:

• sell-through is ahead of plan;
• demand is relatively insensitive to price;
• the product is replenishable or has a stable core lifecycle;
• a later event is known to generate high traffic;
• a deep reduction would create substantial cannibalisation or brand damage.

The decision should be made at the stock-keeping-unit (SKU), location and channel level where data permit. A single chain-wide markdown can be convenient but may destroy value in strong stores while failing to solve the problem in weak ones.

3. Forecasting and Decision Metrics

A robust markdown process begins with a baseline forecast and an explicit forecast of price response. Relevant inputs include:

• historical sales adjusted for stock-outs and prior promotions;
• current on-hand and inbound inventory;
• seasonality, weather, holidays and local events;
• product age, colour, size and style attributes;
• competitor prices and search visibility;
• channel-specific conversion, returns and fulfilment capacity;
• planned marketing and display exposure.

Forecasts should be evaluated using out-of-sample accuracy, not only fitted historical performance. Common measures include mean absolute error (MAE):

where Aᵢ is actual demand, Fᵢ is forecast demand and n is the number of observations. MAE is expressed in units and treats errors symmetrically. Percentage measures can be unstable when actual demand is near zero.

Useful operating metrics include:

A high sell-through rate is not automatically good: selling out too early may indicate an unnecessarily low initial price. Conversely, a low rate may be rational if the retailer intentionally protects margin on a long-lived core item.

4. Markdown Architecture by Retail Context

4.1 Seasonal retail

Seasonal products have a finite selling window. The retailer should establish a markdown calendar before the season, but retain trigger-based flexibility. A useful architecture has three phases: launch and learning, managed acceleration, and terminal clearance. Triggers can combine cumulative sell-through, weeks remaining, stock cover and forecast probability of selling out.

A key principle is option value. Keeping a product at full price preserves the option to earn high margin if demand materialises. Waiting too long, however, leaves insufficient periods to clear inventory. The optimal policy therefore depends on the slope of demand decline and the cost of residual stock.

4.2 Fashion retail

Fashion demand is affected by novelty, social signalling, fit, colour and peer adoption. Aggregated category demand can hide severe SKU-level variation. Markdown decisions should therefore account for size and colour imbalance. A product may have acceptable total sell-through but be left with commercially unattractive sizes.

Fashion retailers commonly use a ladder of moderate reductions followed by a controlled final clearance. They should test whether an early targeted reduction, improved presentation or channel transfer is better than a broad public discount. Full-price availability in premium stores may coexist with lower prices in outlet or digital channels, provided customers do not perceive unfair or misleading discrimination.

4.3 Perishable retail

For food, flowers and other perishables, time-to-expiry is a hard constraint. Dynamic reductions can recover value while reducing waste. The system must incorporate safety, labelling, traceability and cold-chain requirements; no price strategy can justify selling unsafe or non-compliant goods.

A perishable markdown policy often uses age and remaining shelf life as primary variables. The price may fall in stages, but the terminal action can be donation, processing, animal feed, recycling or disposal. Waste cost, disposal cost and social impact should be included in the decision model.

4.4 Capacity-constrained services and retail experiences

In airlines, hotels, events, car rental and appointment retail, capacity is perishable: an unsold seat or room after the date passes cannot be stored. Revenue management uses booking curves, customer segments, cancellation probabilities and fare classes to allocate capacity.

A simple protection-level logic reserves capacity for higher-value future demand. If a unit of capacity has expected high-value demand but a low-value customer requests it, acceptance depends on the expected opportunity cost. In practice, models may use bid prices, nested booking limits or dynamic programming. These methods require reliable demand distributions and controls against overbooking and unfair treatment.

5. Worked Example: Fashion Markdown Decision

An illustrative retailer has 1,000 jackets with a landed cost of $40 each and a current selling price of $100. At the end of week 1, 420 units have sold. The baseline forecast for the remaining four weeks is 80 units per week at $100, giving expected sales of 320 and residual inventory of 260 units.

Management considers reducing the price to $80. Forecast demand at $80 is 140 units per week for four weeks, or 560 units, but only 580 units remain. The expected incremental sales are therefore constrained by available inventory. Assume variable selling cost is $5 per unit, and ignore fixed costs because they do not change between options.

Option A: keep $100. Expected units sold are 320. Contribution on remaining sales is:

(100 − 5) × 320 = $30,400

Expected residual inventory is 260 units. If residual units have zero recovery at season end, expected contribution is $30,400, excluding the already realised sales.

Option B: reduce to $80. Expected units sold are min(560, 580) = 560. Contribution is:

(80 − 5) × 560 = $42,000

The immediate contribution improvement is $11,600. This result does not prove that the markdown is optimal. It assumes the forecast is causal, that the price will remain effective for four weeks, that no competitor response occurs, and that the unsold 20 units have zero value in both options. If the retailer can transfer residual stock to an outlet for $30 net of transfer cost, or if the markdown damages future reference prices, the comparison changes.

A practical decision would test intermediate prices, for example $90, estimate confidence intervals, examine size-level demand and set a review trigger after one week. The example demonstrates why both price and volume must be evaluated.

6. An Applied Case: Northstar Department Stores

The following is a fictional illustrative case.

Background

Northstar sells mid-market apparel through 35 stores and an online channel. Its autumn assortment contains 18,000 units across 600 SKUs. After six weeks, chain-wide sell-through is 48%, below the 60% plan. Performance is uneven: flagship stores are near plan, while smaller stores have excess stock in larger sizes and less popular colours.

Problem

The commercial team proposes a blanket 30% reduction. Finance objects that the reduction may sacrifice margin on strong SKUs. Operations reports rising store congestion and increasing transfer costs. The online team wants a further discount, but merchandising fears channel conflict.

Analysis

The analytics team segments SKUs by sell-through, age, size imbalance, forecast demand and weeks remaining. It identifies three groups:

• Protect: 35% of units, high demand and low stock cover;
• Correct: 45%, moderate demand but excessive cover in selected locations;
• Exit: 20%, aged, low forecast demand and high end-of-season risk.

The team estimates price response using comparable historical markdown events, controlling for store, week and product category. It validates predictions through a small controlled test rather than assuming that historical correlation is causal.

Approach and solution

Northstar applies a differentiated policy. Protect SKUs remain at full price, except for isolated size imbalances. Correct SKUs receive a 15% reduction in weak locations, with inventory transfers from overstocked stores to locations showing stronger demand. Exit SKUs receive 25% initially, followed by a scheduled 40% clearance if the one-week sell-through trigger is not met. Online prices are coordinated with fulfilment economics and a transparent end date.

The governance process requires commercial, finance, operations and legal approval for exceptional reductions. The system records the reason for each markdown, forecast version, expected recovery and post-event result.

Outcome

In the fictional evaluation, end-of-season residual inventory falls by 22% relative to the previous season, while average markdown depth falls from 31% to 24%. Full-price sales in the Protect group rise modestly because availability is preserved. Transfer costs increase, but total contribution improves. The largest improvement comes from earlier intervention on Exit SKUs, not from deeper discounts across the entire assortment.

Learning points

1. Aggregate averages conceal location and attribute-level problems.
2. Markdown depth should follow economic risk, not managerial convenience.
3. Transfers, presentation and channel choice can substitute for price reduction.
4. Controlled tests and post-event measurement are essential.
5. A markdown policy requires cross-functional governance.

7. Technology, AI and Implementation

Modern retail systems can combine point-of-sale data, inventory records, e-commerce behaviour, competitor observations and external signals. Machine-learning models may forecast demand by SKU, location and day, while optimisation engines recommend price, timing and quantity allocation. Computer vision can detect shelf availability or presentation problems that might otherwise be incorrectly treated as price problems.

Technology does not remove managerial responsibility. Models can learn from biased historical markdowns, confuse stock-outs with weak demand, overreact to noisy data or produce recommendations that are impossible to execute. Good practice includes:

• separating causal price-response estimation from simple sales prediction;
• using holdout periods or controlled experiments;
• monitoring forecast error and bias by category and store;
• imposing price floors, legal constraints and brand guardrails;
• documenting overrides and model versions;
• testing recommendations for stability and explainability;
• integrating inventory, pricing, promotion and fulfilment systems.

A practical implementation sequence is: establish clean SKU and inventory data; define objectives and cost conventions; create baseline forecasts; segment products by lifecycle; set triggers and guardrails; pilot in selected categories; measure results; then scale with continuous review.

8. Limitations, Ethics and Governance

Markdown optimisation is vulnerable to several limitations. Demand is endogenous: competitors react, customers wait for discounts, and a markdown may alter traffic and conversion. Historical data may be censored by stock-outs. Returns can reverse apparent sell-through. Price changes can also change product visibility in digital search, making elasticity estimates unstable.

Ethically, retailers should avoid misleading “was/now” reference prices, false scarcity, hidden conditions and discriminatory pricing that uses sensitive personal characteristics. Consumer-protection law may regulate reference prices, expiry labelling, promotional claims and price displays. Perishable markdowns must never compromise safety. Dynamic pricing should be transparent enough for customers to understand material conditions, and governance should include audit trails and escalation procedures.

Managers should distinguish legitimate segmentation—such as different service costs or inventory locations—from exploitative discrimination. They should also consider waste, worker pressures caused by unrealistic clearance targets, and the environmental effects of repeated overbuying and disposal.

Best-Practice Framework

A disciplined markdown programme follows these principles:

1. Diagnose before discounting: check availability, display, assortment, competition and forecast quality.
2. Use lifecycle logic: define launch, review, acceleration and terminal phases.
3. Measure at useful granularity: SKU, location, channel, size and colour where material.
4. Optimise total value: include margin, holding, transfer, disposal, returns and capacity opportunity cost.
5. Use trigger-based decisions: combine sell-through, age, stock cover and time remaining.
6. Protect the brand: use channel architecture and targeted reductions where suitable.
7. Experiment responsibly: test alternatives and evaluate incremental effects.
8. Govern technology: maintain constraints, human review, documentation and fairness checks.
9. Learn after the event: compare forecast, realised sales, margin, residuals and waste.

Chapter Summary

Markdown and clearance are lifecycle decisions that connect pricing with inventory, forecasting and operations. The economically appropriate markdown balances lost unit margin against increased sales, reduced holding cost and lower terminal risk. Sell-through, stock cover, ageing, contribution and recovery rate provide complementary measures; none is sufficient alone. Seasonal and fashion goods require explicit time-window and obsolescence logic, perishables require safety and expiry controls, and capacity-constrained services require allocation of perishable capacity among demand segments. Forecasts should incorporate price response and be tested out of sample or through controlled experiments. AI and optimisation systems can improve scale and consistency, but they require clean data, causal thinking, guardrails and governance. Effective practice is differentiated, trigger-based, cross-functional and ethically transparent.

Key Terms

Markdown: A planned reduction in a product’s selling price.

Clearance: Terminal action to liquidate residual inventory.

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

Inventory ageing: Time inventory has been held or time since receipt.

Stock cover: Estimated duration current inventory can satisfy forecast demand.

Markdown depth: Percentage reduction from a specified reference price.

Contribution: Revenue less relevant variable and decision-dependent costs.

Recovery rate: Value recovered relative to a stated cost or value basis.

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

Obsolescence: Loss of commercial value caused by age, fashion or technology change.

Perishable capacity: Capacity whose value expires if unused at a particular time.

Revenue management: Allocation of limited, perishable capacity to maximise expected value.

Booking curve: Cumulative bookings observed as a service date approaches.

Protection level: Capacity reserved for potentially higher-value future demand.

Forecast bias: Systematic tendency to over-forecast or under-forecast.

GMROI: Gross margin return on inventory investment.

Price floor: Lowest permitted price under commercial or governance rules.

Channel conflict: Harm caused when prices or availability across channels undermine one another.

Reference price: Price against which customers evaluate a current price.

Review Questions

Short Answer Questions

1. Define markdown and distinguish it from clearance.
2. What is sell-through, and why must its denominator be specified?
3. Explain the meaning of stock cover.
4. State two reasons why a retailer might markdown earlier.
5. Why is a high sell-through rate not always evidence of optimal pricing?

Long Answer Questions

1. Explain the economic trade-off between markdown depth, sales volume, holding cost and terminal risk.
2. Compare markdown management for seasonal, fashion and perishable retail.
3. Discuss how revenue-management principles apply to airlines, hotels or appointment-based retail.
4. Evaluate the advantages and limitations of machine-learning markdown recommendations.
5. Design a governance framework for a multi-channel retailer’s markdown programme.

Analytical / Application Questions

1. A retailer has 400 units remaining. At $60 it expects to sell 180 units, while at $48 it expects to sell 300 units. Variable selling cost is $4 per unit and residual units have no value. Which price produces greater expected contribution, and what additional assumptions should be tested?
2. A fashion chain has acceptable total sell-through but large residual inventory in two sizes and three stores. Diagnose possible causes and propose a non-blanket intervention plan.
3. A supermarket is deciding whether to reduce a product two days before expiry or donate it. Develop an evaluation that includes revenue, labour, disposal, food safety, compliance and social impact.

Multiple Choice Questions with Answers

Question 1

A retailer has a reference price of $120 and reduces it to $90. What is the markdown depth?

A. 15%

B. 20%

C. 25%

D. 30%

Correct Answer: C. 25%

Question 2

Which metric estimates how long current inventory will last at forecast demand?

A. Sell-through

B. Stock cover

C. Markdown rate

D. Forecast bias

Correct Answer: B. Stock cover

Question 3

Why can a markdown increase total contribution despite reducing unit margin?

A. It always increases the initial price

B. It can increase units sold and reduce terminal costs

C. It eliminates variable costs

D. It guarantees competitor exit

Correct Answer: B. It can increase units sold and reduce terminal costs

Question 4

In the formula ε = (% change in quantity) ÷ (% change in price), ε represents:

A. Inventory ageing

B. Price elasticity of demand

C. Gross margin return

D. Forecast error

Correct Answer: B. Price elasticity of demand

Question 5

A product has 500 units available and 200 units sold. Using the stated available-inventory convention, sell-through is:

A. 20%

B. 30%

C. 40%

D. 60%

Correct Answer: C. 40%

Question 6

Which retail context has capacity that cannot be stored after the selling date?

A. Core packaged groceries

B. Airline seats

C. Basic household cleaners

D. Long-life stationery

Correct Answer: B. Airline seats

Question 7

Which is the best first response before applying a broad markdown to weak sales?

A. Check whether stock-outs or poor display caused the weakness

B. Reduce every SKU by 50%

C. Ignore competitor prices

D. Delete historical data

Correct Answer: A. Check whether stock-outs or poor display caused the weakness

Question 8

In the illustrative fashion example, what was the expected contribution from selling 560 units at $80 with $5 variable selling cost?

A. $30,400

B. $37,800

C. $42,000

D. $44,800

Correct Answer: C. $42,000

Question 9

Which statement about AI markdown systems is most accurate?

A. They remove the need for governance

B. They always estimate causal price effects correctly

C. They can be affected by biased historical data and stock-outs

D. They make forecasts unnecessary

Correct Answer: C. They can be affected by biased historical data and stock-outs

Question 10

A protection level in revenue management is primarily intended to:

A. Reserve capacity for potentially higher-value future demand

B. Guarantee every customer the lowest price

C. Increase inventory ageing

D. Eliminate cancellation risk

Correct Answer: A. Reserve capacity for potentially higher-value future demand