Phygital’s Prediction Model Enabled a US-based Retail Giant Get Full Control Over Maintaining Inventory Levels
The client is a leading US-based food and grocery retailer with 57+ stores in the Northern USA. They sell various food products, including fresh produce, meat & poultry, dairy, packaged foods, pet food & supplies, and more. Their focus on product quality and any-time-availability of all product types helped them see outstanding growth within five years. The client was looking forward to spreading their footprint across 50 more locations in 2 years to emerge as the number-one retail store in the region.
With fast business growth, the client needed help to balance inventory and demand for items. While overstocking led to wastage, it also reduced profits because the clients had to offer increased discounts to clear the stocks. Likewise, understocking led to the loss of revenue and customer frustration. The client's inability to manage these two issues proved a major stumbling block in their goal of spreading its operations. The other primary challenges were:
- Removing the guesswork from the decision-making process of procuring perishable items.
- Improving customer satisfaction by consistently maintaining a broad assortment of items irrespective of shelf-life
- Estimating inventory level accurately with products of varying shelf-life
- Correctly understanding quality deterioration of perishable items.
- Growing competition from highly targeted online grocery stores operating through dark stores business models.
- Dealing with low-value products with high handling costs and inefficient supply chain operations.
The client sought a long-term and sustainable solution to these persisting problems. They needed an accurate forecasting module to assist them in predicting the demand for perishable and non-perishable items so that they could have complete control over the shelf-life of products. So, they approached Phygital Insights for a solution.
To understand the issues faced by the client, we first built a team of experts comprising data scientists and analysts with hands-on experience in developing solutions for the retail industry. Next, the team coordinated with the floor managers of the retail store and supply vendors to get under the skin of the issues. Next, they analyzed and investigated data sets and summarized their main characteristics to understand how to proceed. Finally, after a detailed study, the team decided to leverage Exploratory Data Analysis to zero in on a solution.
The primary reason for choosing EDA was to understand what data can reveal beyond the hypothesis testing task. Insights into this were critical for a better understanding of data set variables and the relationships between them.
Our team performed EDA on 59 million data transactions collected over 20 months. The disparate data sets were analyzed based on seasonality and trends, stores with the highest sales in a particular duration, geographic location of the stores, items that perished fast, temperature variations in which the items are maintained, etc.
Our solution adopted multiple time series models to consider and analyze all relevant parameters at every level of forecasting. We also created plots in R and Tableau to distill insights from EDA and enhance modeling accuracy. This gave a clear view of the future demand for perishable items and seasonal demand spikes across different locations.
Benefits of Our Solution
Banking on our solution, our clients were able to ensure the following:
- Improved estimation of stocking requirements of food products of varying shelf-life
- Improved sales forecast of specific items to the SKU level leading to reduced stockouts or overstock
- Efficient purchase planning and inventory maintenance
- Improved discount optimization and reduced loss by up to 70%
- High inventory turnover leads to reduced wastage
- Better balancing of customer expectations with marketing insights
Get in touch with us today for all your data science solution needs