an a.i. driven customer credit retrieval engine
hyper-personalisation in customer credit retrieval
For over 10 years, Connect2Collect has been at the forefront of supplying customers in the telecom with personalised and user-friendly debt collection options. Their system can send reminders in the form of e-mails, SMS or VoIP messages. With this, Connect2Collect offers a very user friendly, online communication platform that gives you full control over your communications, interactions and transactions.
“In this case it was of high importance to not only focus on the technology, but also on the business models”
ABOUT THE PROJECT
Wolfpack has closely collaborated with Connect2Collect to develop the software behind their Connect2Collect and Pulse Connected Engine services. In this case it was of high importance to not only focus on the technology, but also on the business models behind their services. By following their business model and implementing real time machine learning algorithms, the platform is now able to collect late payments much more effective than before. Next to that, the platform was designed to interact with numerous external API’s, which means that connecting with new systems is now as simple as installing a new laptop.
“Onboarding new clients or adding external API’s is now possible in a day or two”
Assessing the legacy IT infrastructure is the most important part of the project, as it not only allows us to determine the scope of the project, but also allows us to define opportunities for innovative solutions. During this phase, two major goals were defined, based on Connect2Collect’s key performance indicators. First was to allow the system to scale with the growing number of customers and traffic, and secondly, to use smart algorithms and machine learning to optimise their customer conversion. Both developments would allow Connect2Collect to shift their business towards a more optimised model.
To allow the system to be able to scale along with a growing number of customers and traffic, we developed a complete new back-end, utilising a microserviced system. This allows the platform to interact with numerous external parties and their API’s, without having to compromise in technical features, or programming languages. Onboarding new clients or adding external API’s for new communication options is now possible in a day or two.
In order to optimise customer conversion, we developed several machine learning models that were applied to the data of Connect2Collect’s customers. With this, it became possible to use predictive analytics to determine which types of customers would respond best to which kind of message. Factors included time of day, channel, structure, tone of voice, layout, and more. With this, it is now possible to predict for 95% of the clients if they will pay or not, and even increase their customer satisfaction.
In this project we implemented several business-oriented IT solutions in order to transform an older monolithic system into a robust, extendable and intelligent system. Firstly, a completely new micro-serviced back-end was developed. This allows external services to be easily and quickly connected to the system, which creates much more flexibility when adding new clients or customers.Using these Machine Learning models the engine can predict several factors that optimize the conversion and customer interaction. This factors include payment probability, optimal message send time and date and message type and tone. With this, it is now possible to predict for 95% of the clients if they will pay or not, and even increase their customer satisfaction.