Markdown optimization
to clear stock and maximize margin

Markdowns are costly. According to the survey, annual revenue loss due to markdowns comprises up to 12% of total sales. However, the alternative - storing seasonal items or writing off expired goods - comes at an even higher price. While markdowns are inevitable, they still can - and should - be an essential tool to implement a retail strategy by managing inventory and maximizing margin.

Move products before the expiration date
The inevitable nature of markdowns is especially true for grocery retailers who deal with perishable goods. Once expired, these goods are written-off and thrown-away. In fact, perishable goods account for 70% of food wasted worldwide, with dairy, meats, fruits, and vegetables being among the categories that comprise most of the spoilage due to the shelf life of several days. Thus for grocers, the most common reason to markdown is the necessity to move products before the expiration date.


Clear stock during assortment change
Turnover is among the key parameters defining a retailer's overall success. Inventory that is not selling is a stagnant capital retailer cannot use. It incurs storage costs and occupies valuable shelf-space that otherwise can be used for items that sell faster and at a full price. To make inventory moving, markdowns can be used. For example, they allow clearing stock to make room for new product lines during assortment change.


Increase sell-through of seasonal items before the season ends
Seasonality in grocery retail often concerns fresh produce (fruits and vegetables) and foods specific for certain holidays and occasions. Easter eggs significantly lose market value after the holiday is over. To sell easter eggs before the Holy Week ends, retailers have to markdown. Thus a third common cause for markdowns is seasonality, which forces retailers to clear their stock before the season ends.

Markdowns turn up to be an essential tool in moving inventory and managing margin. However, more often than not, it's not used to its full potential. To know what, when and how much to markdown, retailers usually use a set of predefined rules. An example of typical rules may be: if the product expires in N days, cut the price by X%; or if the product doesn't sell for N days, cut the price by X%.

Fresh milk with a nearing best-before date sells through with a 30%-discount, but would it at 25%? The optimal price cut will differ between various categories, seasons, store locations, and the markdown period. Discount too little, and you get overstocked; discount too much, and you cut already-thin margin.

With abundant data and computation power at the disposal of modern retail chains, there is a better way to optimize the markdown process than relying solely on the rules - machine learning (ML).

Moreover, ML-based markdown optimization automates high-intense tasks associated with markdown planning and calculating the optimal price cut rate. Depending on the markdown policy in effect - clearing stock or maximizing margin -, a potential decrease in write-offs due to shelf-life expiration can reach up to 50% while margin can increase by 2%. While clearing stock may result in a temporary decrease in margin, it is also an investment in customer loyalty: markdowns can be a powerful incentive for choosing a grocery store over competitors'. As long as customers buy ordinary priced goods along with the marked-down once, this behavior is beneficial for both customers and retailers.

DSLab's markdown optimization service integrates into demand forecasting and pricing processes and recommends the best price for each SKU identified for a markdown on a store level:

1. When the markdown process is triggered by one of the rules on the retailer's side, the service forecasts potential overstock, identifies sales gain to avoid it, and recommends the optimal price.
2. These recommendations are automatically adjusted to the constraints of the markdown strategy (i.e. minimum and maximum price changes) and retailer's KPIs for the campaign (i.e. maximizing margin or clearing stock by the termination date).
3. To optimize markdown actions, DSLab's service utilizes machine learning technologies to model price elasticity and future demand. We model price elasticity based on historical data from past campaigns and price changes for a given product or product category. As markdowns are typically caused by demand forecasting errors, using the same demand forecasting model to calculate markdowns can be erroneous. To avoid it, we use another model to calculate potential overstocks and optimize markdowns.
4. Markdown-related data are then integrated into the forecasting system with a special label to avoid corrupting the primary demand forecasting model with abnormally high sales.

DSLab's markdown optimization service easily integrates into retailers' demand forecasting and pricing systems via API. It adjusts to retailer's KPIs and delivers measurable results such as increased margin or decreased write-offs.

Markdowns are inevitable, whether we like them or not. However, with the right technology in place, grocers can utilize them as an essential and highly-effective tool for implementing retail strategy and delivering business value through managing inventory and maximizing margin.