Willis Towers Watson has updated its Emblem predictive modelling software with a technique that allows users to model complex customer behaviours involving multiple possible outcomes quickly, and with greater pricing accuracy.
The company claims insurers can fit models to data sets in seconds thanks to the update. It is 20% faster than the previous version on rich datasets, enabling “significantly reduced” decision cycle times.
“This substantial improvement helps companies more accurately model complex customer behaviours with multiple outcomes. The ability to predict the purchasing behaviour of customers presented with such propositions offers insurers a real commercial edge when operating in competitive markets,” said David Ovenden, global director, pricing product claims and underwriting.
Emblem had its previous major release in 2016, when the firm introduced multinomial modelling. It enabled users to detect previously unnoticed patterns in insurance claims and customer behaviour experience within a single modelling framework.
Firms who use multinomial modelling are able to predict the probabilities of a set of related events, and also the probability none of those events occur. According to the firm, its real-world uses include:
- Multi-brand new business demand
- Multi-product demand offering
- Add-on package purchasing
- Loan survival and delinquency
- Claims tiering
“With competition intensifying, changes in regulation and distribution, and the ongoing redefinition of consumer expectations, we are seeing a clear and widespread focus on pricing sophistication and effective customer management. As part of that, interpretable and implementable analysis is becoming essential for product management,” Ovenden said.
Related stories:
ILS market prepared to respond following disasters – Willis Towers Watson
Willis Towers Watson appoints leader for ‘high-potential’ region
.jpg)
