Asset Publisher

Breadcrumb

Predictive Analytics: Propensity to Buy Model

Machine Learning Project Delivers Data-Driven Capabilities to Food Service Drivers

Solution: Analytics, Case Study, Analytics Case Study

Industry: All Industry, Food & Beverage

Project: Machine Learning Project Delivers Data-Driven Capabilities to Food Service Drivers
Client: Food Delivery Service Company

For nearly 70 years, this family-owned-and-operated company has worked hard to bring personal service and delicious meals to homes and communities across America. They are the largest direct-to-consumer frozen food delivery service in the U.S., offering food delivery to 48 states and employ 4,150 people.

Challenge: Delivery Drivers Sales Calls Were Not Optimized

Food services drivers not only deliver orders but also function as salespeople, stopping at potential customers when they are in the neighborhood. Drivers would often visit customers that were either not at home or who were not interested in purchasing anything at that time. This resulted in sales calls which did not result in any revenue.  The company challenged Dunn Solutions to identify a data-driven process for food delivery that will increase sales and minimize the cost of delivery based on visiting the customers that have the highest chance of purchasing to make the drivers and company more profitable. In addition, we were asked to create real-time reports and a method of running ad-hoc reports.  

Solution: Dunn Solutions Creates a Machine Learning Model to Guide Drives and Increase Sales

Dunn Solutions conducted exploratory data analysis provided by the company on their previous customers and sales to better understand how machine learning could help. Next, we developed a model that assigns a “probability of making a purchase” score to all customers of interest to use for sales/routing optimization efforts. This “Propensity to Buy” machine learning model predicts which customers are ready to buy at a particular date and time and prioritizes them based on the driver’s routes.  

Result: Propensity Scores Guide Driver Sales Activities and Yield Higher Revenues

Using our Propensity to Buy Model to plan the routing, customers with low scores are de-prioritized on the route. The driver can re-route based on the model, telling them where they have the highest likelihood of a sale. Our solution: 

  • Increased driver efficiency 
  • Saved time on wasted visits 
  • Saved mileage and wear and tear on the delivery vehicle 
  • Strengthened relationships with loyal customers 
  • Generated 1% more revenue per day with no additional driver or marketing expense

As a result of the modelling, the food delivery service company is able to create a more boutique, customized approach to food delivery. They can provide a personalized, hands-on experience for their most valued customers with higher revenues and lower staffing.