Some industries like Transportation or Energy have long ago introduced dynamic pricing: the ability to change the price of products or services rapidly and automatically. In the insurance sector, successful comparison web portals and new customer behaviours are currently confronting even well-established and renowned insurances with the necessity to dynamize their prices, unless they accept to lose market shares and reduce their profit margin.
Moving from legacy data analysis to a modern dynamic pricing based on more volatile data, like competitor prices or customer behaviour is not an obvious task, from the point of view of the mathematics, infrastructure, software, team culture.
airsecoma GmbH’s focused skills in data analytics, machine learning and software architecture is a valuable partner to accompany your migration to a state-of-the-art data first strategy.
Krier Information GmbH, German specialist for insurance industry digitalization, engaged airsecoma GmbH co-founder and consultants led by Jerome Rougnon-Glasson to assist it in developing the dynamic pricing solution at the third largest German insurance. The airsecoma consultants worked on the system architecture of the pricing engine, allowing Krier Information GmbH to focus on project management and domain issues, specific to insurance.
The result of the cooperation is a cutting-edge dedicated pricing engine processing millions of quotes daily in a few milliseconds by evaluating highly complex pricing models. The engine is combined with a live analytics stack allowing live monitoring, post-operation analysis and live training of machine learning models.
Assisted by the Technical University of Munich, airsecoma GmbH team members, led by co-founder Jerome Rougnon-Glasson, selected a dozen of algorithms and approaches. The team implemented, fine-tuned, and evaluated them. By applying state-of-the-art methodology of machine learning, the team was able to iteratively eliminate many of them, that did not reach the expectations. It then selected the best prediction models, fine-tuned them again to outperform human trader performance. Finally, it provided conventional engineering services to integrate the algorithms into the company trading system, to assist traders in making decisive choices as well as enabling full trading automation.