Customized restaurant recommendations to build customer loyalty
By Centre de recherche informatique de Montréal (CRIM) | Published on February 17, 2022
Problem and objectives
Problem
- UEAT had a digital solution without customized recommendations.
Objectives
- Customized menu based on the customer and external factors;
- Creation of an AI-based competitive advantage for UEAT;
- Increase in number and value of online orders for restaurant owners;
- Maximizing customer retention.
Proposed solution
Solution
The Computer Research Institute of Montréal (CRIM) is a non-profit organization whose goal is to help organizations with research projects tailored to their needs. The firm UEAT retained CRIM’s services to customize the menus of restaurant owners who use their platform. Together with UEAT, CRIM developed a recommendations system powered by Artificial Intelligence (AI).
Principle
While major restaurant chains invest considerable sums in developing digital solutions, it can be difficult for small restaurants to do the same. In most cases, small restaurants use a static menu in PDF form. To counter this, UEAT’s objective was to provide various restaurant industry players with an online ordering system. To customize the different offerings proposed to customers, it was imperative for UEAT to look at AI techniques with their partner, CRIM.
The recommendation system developed by CRIM is based on:
- The customer’s profile, including order history, preferences, etc.;
- External factors, such as weather, seasonality, events such as hockey games, etc.
User experience
By browsing a restaurant’s website, customers can select the items they want to order through the platform developed by UEAT. The algorithm then suggests items to consumers based on their preferences and external elements such as the weather or season. The suggestions increase the conversion of visits into transactions and help visitors place their orders more quickly.
Role of AI
Use of AI improves suggestions by providing recommendations tailored to each user, which would have been impossible to do manually due to UEAT’s extensive service area. Restaurant owners who use the platform can also focus on the management aspects of their establishment and avoid having to change their menu layout.
In developing the algorithm, CRIM used a variety of techniques, including:
- Field-aware Factorization Machines (FFM);
- Clustering;
- Machine learning;
- Deep learning.
Impacts
Main outcomes
Impact on UEAT:
- Creating a competitive advantage through the AI tool; . Development of AI knowledge through the project with CRIM.
Impacts on restaurant owners:
- Sales five to six times higher for businesses using UEAT products. Optimization of consumers’ time to choose their meal.
Impacts on CRIM:
- Increase in experience associated with recommendations customized by AI.
Economic development
- Increase in restaurants’ sales;
- Increase from 7 to 65 employees at UEAT over a three-year period.
Challenges takled
- Management of customer’s expectations and understanding of what can be done with AI;
- Working with a limited budget and staff;
- Maintaining an iterative structure in the development of the solution.
Conditions for success
Team mobilised
A total of four people worked on the project. The CRIM AI project team was made up of:
- Data engineers;
- Data scientists;
- A project manager;
- Data analysts.
Collaborations
The following collaborations contributed to the success of the project:
- Major support from UEAT management;
- Contribution, as needed, from software developers or architects external to the project within CRIM;
- Creation of a sharing environment with VMware that enables collaboration between CRIM and UEAT.
- Involvement of CRIM’s ethics committee.
Project stages
Interactive visit through the Industrial Research Assistance Program (IRAP) – Problems explained by UEAT and development of a solution.
Development of a proof of concept;
Development of the solution in collaboration with the company.
Solution managed by UEAT (complete independence).