
Communications, Media & Telecom - Building a Near Real-Time Recommendation System
Communications, Media & Telecom
As part of the cloud integration, an algorithm was developed to determine the best suggestions for each customer and to integrate them across various channels, leading to a significant increase in conversion rates.
Initial situation
The customer was confronted with a variety of products and tariffs that had grown over decades. A systematic recording of the features, including costs, was no longer possible. Additionally, the offers were not systematically presented to customers but managed through various channels with different logics.
Architecture
The architecture is built on AWS, using technologies such as AWS Sagemaker, AWS Glue, and Model Management.
Generated benefits
Conversion rates have increased by a double-digit percentage across all sales channels.
Services accomplished by synvert
As part of the cloud integration, synvert rethought the case from the ground up. Due to the changing tariff portfolio that brokers will continue to face in the future, traditional algorithms for recommendation systems are no longer effective. Instead, synvert developed an algorithm through mathematical modeling that determines the best possible suggestions for each individual customer. With the help of intelligent parallelization in the cloud, this is also possible for many potential products. Different business logics can also be implemented for various sales channels. The suggestions are then fed into the various target systems via an API.