Assortment recommendation engine maps the right products with outlets leading to 14% sales growth
Sigmoid developed an assortment optimization solution with customized prediction and recommendation models using Databricks to help a leading alcoholic beverages company improve customer experience by placing the right products at the right outlets and drive sales.
The client with a diverse product portfolio, wanted to optimize product assortment across all the off-premise outlets such as liquor, convenience and retail stores, and on-premise outlets such as bars and restaurants. But, personalizing assortments for each outlet based on historical sales data and distributor inputs proved to be complex and time-consuming. Thereby, the client sought a data-driven solution that would help them enhance product placement, align assortments with consumer preferences, and maximize revenue opportunities.
Sigmoid created an assortment based solution using advanced analytics and ML algorithms that recommended the right product-outlet combination and estimated the sales potential for each product at an individual outlet. The solution processed large volumes of data from various sources including product, demographic, product performance (sales), and outlet data in the cloud data mart on Azure. Two types of models were created on Databricks to achieve the objectives; recommendation model predicted the likelihood of a product to be recommended in an outlet and opportunity prediction model predicted the sales opportunity at a product-outlet level. The product-outlet level data was further fed into the Alteryx-enabled app and dashboards that were accessed by distribution and commercial planning teams.
The solution helped the client capture the nuances and dynamics of the market, leading to more accurate recommendations and informed decision-making. Overall, the client was able to realize portfolio growth, incremental revenue and gain significant market share.
for existing product x outlet combination at national level
14% YoY growth
observed for every new recommended product x outlet combination
between projections and actual sales