The retail industry is rich with data. There is so much data, and to make use of it, we need to use data mining and analytics to drive meaningful insights that can be put to some good use. This blog will review how category managers and supply planners can use cluster analysis to make smarter decisions and grow sales.
Retailers can apply cluster analysis techniques in different ways to find groups of customers, products, stores, or suppliers that behave similarly. Let’s look at the four most common ways cluster analysis is used for category management and supply chain planning decisions.
ABC analysis creates product segments by grouping products with similar sales volume or purchase frequency to enable category managers to focus on what matters most. Traditionally, this has been done by ranking the products based on sales performance.
Now, we can do better than that by using advanced analytics, which can be extended to include more dimensions, such as trends, and provide further insights. Looking at the change in sales as an additional dimension proves helpful in identifying SKUs that may need attention. There might be some diamonds in the rough with sales but lack high sales growth, or some of the top sellers might be on a decline.
Automating this process and looking for exceptions for retailers and e-commerce companies benefit category managers and facilitate faster decision-making.
These ABC segments can also be used for setting up service level targets or availability targets which then determine the safety stocks for ensuring there are enough stocks to meet the fluctuating demand for those groups of products that are critical to business. Instead of applying broad brush service level targets across the board, focusing efforts and inventory investment in segments of products that have a significant impact on sales and are on a growth trend will improve inventory productivity and return on investment.
Price index and price elasticity are useful metrics on their own, and a combination of these can help determine the right price point to maximize revenue and profit. The price index for a product tells us the price position of a seller relative to the competitors’ prices, requiring continuous tracking of competitor prices.
Then comes the juicy part, establishing a relationship between the change in the price index and the change in demand for a given product or groups of products; this is Price Index Elasticity. This is where advanced analytics and AI/ML techniques help.
When we see fluctuations in demand that are not explained by a seller’s price actions or promotional offers, it is always good to look out at the market and see if the competitors’ actions are causing these changes.
In E-commerce, where prices constantly change, clustering techniques can be used to find groups of products more sensitive to price index changes. Those would be the products to focus on to avoid losing market share to competitor price actions.
There are different ways of doing store cluster analysis. One way is to create a store sales profile using different product groups’ sales contributions. Comparison of these profiles then gives us store clusters with similar demand, and based on these insights, assortment plans can be developed to address the difference. Another way is to cluster stores by looking at a specific category’s inventory productivity or sales performance. Since there are often differences in store sizes, a metric representing the size can be added as well for a two-dimensional clustering approach. Average sales price or price elasticity can be a third dimension to identify stores with more price-sensitive customers.
Computationally, there’s no limit on how many dimensions can be used for clustering stores. However, using metrics that can be easily tracked and updated is recommended. As consumer behaviors change continuously, clusters should be updated periodically, at least once or twice a year, or even more frequently depending on the nature of the business and the level of flexibility for assortment changes. Since fashion retailers are constantly bringing new products, they can apply this kind of clustering for each collection.
With our current technologies at our fingertips, it is time to uncover opportunities for growing topline sales and addressing operational challenges that are roadblocks to success. Clustering techniques enable category managers and supply planners to find these opportunities and challenges quickly. Happy clustering!
Following her consulting tenure with Silicon-valley based software companies, she brought her Advanced Analytics and Business Process Engineering experience to Apparel Retail where she worked at Gap Inc. and Cache, where she led Merchandise Planning and Distribution teams to advance analytical capabilities in Forecasting, Assortment Planning and Price Optimization to improve sales and profitability.
Asena holds B.S. and Master of Engineering degrees in Operations Research and Industrial Engineering from Cornell University. She is one of the contributing authors in the Oxford Handbook of Price Management and has been a guest lecturer at San Francisco State and Columbia University business school on multiple occasions.
The post Four Ways of Using Cluster Analysis in Retail Management to Improve Sales appeared first on Logistics Viewpoints.