In our AI webinar, Head of Data Mirko Schuh explained how artificial intelligence can support receivables management. Below we present a summary of the main themes covered in the webinar:
The major difference between a classic algorithm and an artificial intelligence lies in its ability to learn. As a self-learning algorithm, an AI has the capacity to use statistical techniques to detect patterns in data sets by itself – you’ve probably heard people talk about “machine learning”.
When trained on large datasets, an AI can acquire specific sets of knowledge through machine learning. This might be in the field of image recognition – for example, the high success rate at which an AI can detect abnormalities on MRI images. Or, in the field of receivables management.
For AI to be used, processes need to be digital. After all, it’s through digitization that you generate the data required for the learning process. In the field of receivables management, this calls for digitized dunning processes, incorporating automated payment reminders sent via email or text message and interactive landing pages with digital payment and feedback options.
Once a digital dunning process has been set up, AI can be used. At collectAI, we utilise a contextual bandits algorithm programmed for reinforcement learning (to find out more, view the recording of the webinar).
The AI system interacts with the digital communications strategy and makes decisions about things like when and via which communication channel payment reminders should be sent in order to produce the desired effect. Depending on the customer’s response (i.e. whether or not a payment is made), the AI system recognises how successful its decision was and learns accordingly.
For a certain customer segment an email sent at 4 pm may generate more payments whereas for another customer segment it may be better to send a text message at 9 am in the morning. This is the great strength of AI. In statistically complex scenarios with large data sets, AI recognises patters and translates them into granularity-driven decisions and measurable success.
