Dynamic pricing and markdown optimization of perishable products in retail stores

Abstract

The Indian Retail sector is estimated to be reaching a size of US $850 billion by 2020, of which organized retail is expected to grow at 25% and reach a size of US $200 billion1. Organized retail stores have started selling everything from groceries, fruits which have a very short life span to computers and advanced electronic devices with a longer time frame. For products with shorter life span there are multiple challenges. Because of the short life span, perishable products needs to be sold within their life span else there will be a loss to the firm. Price reductions are often used as a mechanism to push sales of such products. Reliance retail is one such firm which is facing challenges in this front. Reliance retails sells groceries, milk and fruits which have limited lifespan and often results in dumping of such products which reduces the profit. A right pricing strategy for perishable products will help the company in increasing revenues. Pricing such products dynamically based on the pending shelf life will maximize the expected profit and also reduce dumping of the products. The project develops a data based approach to maximize profits of such perishable products. The project focuses on analysing banana, which is one of the largest dumped item in terms of value and also quantity for Reliance Retail. The focus is to increase the profitability by finding the optimal pricing for various shelf life value of bananas. With this aim a regression model is developed with demand as an independent variable and customer turnover, shelf life of banana, price of banana and if the day is a weekend or not as dependent variables. From the model, it is found that demand increases with increase in customers and shelf life and also on weekends. It also decreases with the increase in price which indicates that banana is a price sensitive item. Once the demand is formulated, the objective is to find the price which maximizes the profit. For this, a dynamic programming approach is used. In the dynamic programming method, optimal pricing for each day is calculated based on the model and also on the actual sales. The end users of the model are store level employees who do not have much sophistication in the data analytics area and hence a tool was developed so as they input the necessary value and the prices are calculated based on the inputs.