Markdown Optimization Case Study

Markdown Strategy

Improving margin by establishing better markdown management through rule-based framework, training and reporting


The Client

A $15B+ specialty retailer, of hardlines and softlines through multiple channels, store formats and brands. Focus was on the softlines business.


The Challenge

The retailer was transforming its softlines business and a large component of the expected business benefit would come from shifting clearance margin to regular/promo, and increasing the clearance margin rate through better markdown management. Challenges in the current markdown process included:
− No seasonality visibility
− Limited markdown rules in place
− Markdowns taken at national level
− No integration to merchandising or marketing calendars
− No visibility or control at item level after first markdown
− No 'what if' capability


The Parker Avery Solution

The Parker Avery Group helped the client address the challenges identified in the current state by recommending six initiatives:
1) Establish a markdown rules framework by product type
2) Add markdown rules to the assortment planning process
3) Improve reporting to provide visibility to markdown performance
4) Implement rule-based, localized markdown capabilities
5) Establish a markdown training program for merchants and planners
6) Implement markdown optimization


The Result

As a result of the project, the client expects to achieve the following benefits,
• +3% shift from clearance to reg/promo margin
• +1.3% clearance selling margin rate
• Maximized sales
• Improved sell-thru


Download PDF Version

Return to previous page

We would advise the following generic steps in order to create an intelligent markdown pricing solution:
  • Understand the existing markdown pricing process.

  • Carry out preliminary analysis to examine levels and patterns of waste losses currently being incurred.

  • Calculate PEs for products being marked down.

  • Create an alternative markdown pricing strategy designed to reduce the losses.

  • Simulate the impact of the pricing strategy and measure its benefits.

  • Carry out a live test to see how customers react to the change in markdown pricing.

A brief explanation of each step follows below. Bear in mind that the detailed approach will depend on industry and the available data.
  • Understand the existing markdown pricing process

For intelligent markdown pricing to add value, it needs to fit into the existing markdown process — therefore, a detailed understanding of the current process and markdown strategy is an essential first stage. For example, as identified above, retailers selling perishable products may have to carry out daily checks to identify the items reaching the end of their selling periods, quantify the numbers of units that need to be marked down, calculate reduced prices and apply these to the items. This process can be highly complex and staff intensive.

The performance metrics that are used to measure each store manager may include a target on product wastage (due to reaching the end of the ‘sell-by’ date). If so, it is important to identify and understand the metric — for a new markdown pricing strategy to be successful, it will have to support (or at least not conflict with) the manager's performance target.

The existing operational process needs to be fully understood in order to introduce a seamless intelligent markdown process that avoids creating additional workload or unnecessary complications. Although running the analyses might seem difficult, changing existing business processes may prove to be at least as hard. And that is where the money is made.
  • Preliminary analysis to examine business questions concerning waste losses

The initial analysis should quantify the levels of losses and understand how these vary by factors such as product price, day of week and outlet type. If granular data are available, then more detailed business questions can be answered, as discussed above. Where does the business think they are losing money? How does this line up with the data? Losses are not always greatest where they appear to be, and therefore the preliminary analysis can yield unexpected findings and new insights. Data analysts can drive change here by putting together compelling business cases. Providing insight alongside ‘bare’ predictions fosters adoption of intelligent markdown pricing, and may spur innovations in current (store operation) practices.
  • Calculate PEs for products to be marked down

Historic sales and price data are used, together with other attributes (such as seasonality) to calculate the PE coefficient of each product. The PE coefficient measures the slope of the relationship between price and demand, defined and illustrated hypothetically in Figure 2.

Since demand usually decreases as price increases, the PE invariably has a negative value (see, eg Tellis7). If the PE is large in size (eg −2 or less), then a small price reduction will greatly increase the demand and therefore should be sufficient to clear the stock to be sold. On the other hand, if the PE has a smaller size (eg between 0 and −1), then the product is less sensitive to price, implying that a deeper price reduction will be required.

For the most accurate and valid calculations of these demand curves, historic sales as affected by price changes (markdowns) are required. This typically involves analyzing data for each day on which markdowns took place, modelling the relationship between price reduction and quantity sold. Given the central role these elasticities play, great care should be exercised to gather as much useful data as possible, and to ensure their validity.

In some cases, reducing the price of Product A will impact sales of Products B and C, which either compete with Product A or are complementary items. This situation is common in supermarket retailing — for example, reducing the price of apples will increase their sales but may have the effect of reducing purchases of oranges. In this case, it is also necessary to calculate the ‘cross-PE’, that is, the relationship between price of apples and sales of oranges, and to take account of this relationship when deciding the markdown reduction on apples.
  • Create an alternative markdown pricing strategy designed to reduce the losses

Putting the PEs together with the stock levels and forecast sales, the analyst arrives at a markdown pricing strategy for each product. This will include both the markdown discount and the optimal time at which the reduction(s) should be made. An understanding of current systems is essential in order to ensure that the new markdown strategy can effectively be deployed within the existing business process — for example, there is no point in producing a strategy that requires different fresh food products to be marked down at different times of the trading day, because this would be impractical to put into practice. This demonstrates the intricate interplay between analytics, people and process.
  • Simulate the impact of the pricing strategy and measure its benefits

The analyst applies the new markdown strategy to a sample of recent trading days, using the PE models to predict each product's sales at its reduced price. The following steps are involved:
  • The new markdown pricing strategy is applied, in order to reprice each product, for each day on which markdowns were required.

  • The effect on sales is predicted, by applying the PE models.

  • This gives a predicted outcome for that day, that is, the quantity sold (at reduced price), the quantity unsold (for wastage) and the resulting financial loss.

Needless to say that to make this simulation valid, the sample days should not have been used for building the model!

The overall return for each product category can then be derived and hence it is possible to calculate the loss saving to be made by introducing the proposed intelligent markdown pricing strategy.
  • Carry out a live test to see how customers react to the change in markdown pricing

Although the simulation results will predict the return from intelligent markdown pricing using the data alone, the only way to measure its actual success is to carry out a live test using a sample of customers or a small number of stores. The proof of the pudding is in the eating, and management should not attempt to bypass the live test in order to save time — if their need to implement a new price reduction strategy is that great, then think how much more deeply in trouble they will be if they bypass the test and the new strategy fails.

Critically, the test should include a matched control sample, for comparison purposes. Matching is preferred over one large random sample of test and control stores, in order to increase the statistical power of this test. Matching implies finding pairs of stores that are as comparable as possible, and then randomly selecting which of each pair eventually becomes the test store. In ‘ordinary’ sampling one would randomly select test and control stores, but this might lead to large stores being included as test store, and smaller ones for the control group (by chance). Matching reduces this between-group variance. It needs to be carefully designed in order to avoid biasing the results in any way. The design merits care, because a more valid test design leads to more compelling business cases.

Two important reasons to include live tests are firstly to gather the most valid empirical evidence to convince all stakeholders of the value of intelligent markdown pricing, and secondly to experience in a realistic setting where adoption of analytics will affect the operational processes currently in place.

0 thoughts on “Markdown Optimization Case Study”

    -->

Leave a Comment

Your email address will not be published. Required fields are marked *