NBO & Marketing Optimization
- Customer
- Kompanion Bank
- Project manager on the customer side
-
- IT Provider
- Databorn KZ
- Year of project completion
- 2024
- Project timeline
- June, 2024 - September, 2024
- Project scope
- 3200 man-hours
- Goals
-
- Analyze the current marketing offer generation process and infrastructure for modeling.
- Develop a Terms of Reference and Roadmap for implementing the NBO product in a bank
- Build the first ML propensity model for the Cards department.
- Conduct the first marketing campaign with active communications.
- Boost sales through the implementation of the ML model.
- Project Results
-
- The first propensity model for cards was built and tested.
- The conversion rate to card orders increased by 500%.
The uniqueness of the project
The project's uniqueness is that it is the pathbreaking project in the Kyrgyz market using artificial intelligence to build NBOs in the bank.
- Used software
-
The project is fully implemented using a free open-source stack:
Python,
SQL.
- Difficulty of implementation
-
During the project implementation it was necessary to establish business processes for the application of artificial intelligence. Since no aggregated customer features were available, It was important to develop a Feature Store and collect the necessary data for effective modeling. In a short period of time develop an MVP product, test it in a real marketing campaign, demonstrate an increase in sales.
- Project Description
-
Kompanion Bank successfully implemented AI technologies for the sales of its banking products. The project was implemented with the support of IT partner Databorn and was fully realized using an open-source technology stack.
Analysts developed the MVP of the Next Best Offer model, and implemented a Feature Store to accelerate experiment testing. The solution allowed the bank to establish a digital sales process and achieve 5 times increase in the conversion rate from application to purchase.
The method is based on the use of machine learning models and mathematical optimization to build the process of interaction with customers in the most efficient way. The solution selects the best communication channel and the most suitable product for each customer. Moreover, the implemented technologies help optimize sales KPIs and take into account business constraints when interacting with customers and selling them the bank's products.
As part of the project, the analysts developed a roadmap for further development of the technological solution, which will help achieve better KPI results in sales. In addition, due to the development of FeatureStore and MLOps technologies, the deployment of AI models into production will be significantly streamlined.
- Project geography
- Kyrgyz Republic