About the Customer
The client is Hongkong based one of the largest retailers, having 1852 stores spread across the globe offering various products. Their store varies from small shop-in-shop, mid-sized high street store to the larger store in the mall and uses a variety of channels to serve the needs of the customers.
They employ 300,000 associates approximately and aim to ensure each store serves the customer’s need-based in that particular territory.
As the company has thousands of stores across the globe, the company wanted to create store clustering based on the category, country and average sales per square foot. They wanted to divide stores into 12 to 15 different groups so that they can build the same marketing strategy.
Some of the challenges include:
- Need to consolidate the last 3 year’s retailer’s data sources (format like MySQL, CSV, etc) of sales transactions into a single form of data.
- As there was a large number of product sets, it was difficult to derive the common category of the products.
- Wanted to enable more localization of seasonal products
After analyzing the challenges and requirements of the client, SPEC INDIA designed a solution for store clustering that can help to give an edge to the business.
- 1852 stores divided into 12 large clusters based on the category, country as well as sales of each category
- Developed ETL jobs to transfer the different forms of the data into the MySQL database
- Developed data models and transformed the data from staging to data marts.
- Derived Additional Variables
- Explored data sets by generating the descriptive statistics
- Scaled and Weight the variables based on business team inputs.
- Perform the K-Mean Clustering for 10 to 20 clusters groups
- Profiled the cluster and derived the observations for each cluster
Tools & Technologies
By applying the solution following benefits/business details has been derived which helps the client’s business in many ways
- 12 Outliers stores detected having less space, but their sales went higher than the average sales.
- 17% of the stores have a higher than average sales.
- 30% of the stores have significantly higher tobacco and alcohol sales
- 12 marketing strategies were defined based on the observations and deployed across all the 1852 stores.
- Strategies were monitored for over 3 months and sales were increased by 7%.
- Increased foot traffic also resulted in greater opportunities for impulse buying.
- Speed and Efficiency was increased as retailers could create a localized assortment for multiple locations