This model allowed the client to determine why certain customers were more likely to buy certain products.
Helped the client better understand customer needs and desires and the extent to which they were fulfilled by the product.
Improve the customer experience by pre-empting their requirements, streamlining the onboarding processes and resolving any potential issues that can be predicted.
Have more control and insight into potential compliance issues (e.g. KYC, sanctioned customers, money laundering, or mis-selling).
We met with the client to understand their needs and agreed on key data sources, addressing ethical, governance, and security concerns. Using Python, we built and trained a machine learning model on customer purchase history to predict buying propensity. After testing, the model predicted potential customer behaviour on the client’s online quote system, aiding sales and revenue growth. The model was designed for easy integration into existing sales processes, providing augmented intelligence. We also identified options for automated responses in the sales process, such as customer follow-ups.
Crowe’s client, an insurance company, needed to predict the behaviour of potential customers who were engaging through their online quote system. This in turn, could enable them to maximise growth and reduce dropouts in the sales cycle.
Approach
Challenge
Design and deployment of a bespoke machine learning model to predict customer propensity to buy.
We built a model for the UK entity of this global insurer to determine the factors that might influence sales and predict which customers were more likely to buy their product.
Case study
How we have helped our clients
Outcome and benefits
Crowe’s analysis demonstrated immediately the level of inconsistency in existing categories.
It also highlighted opportunities to simplify and eliminate the level of duplication and overlap of different events within one of the existing categories.
It demonstrated how predictive analysis (forecasting) could be used to categorise the risk events more effectively going forward.
We helped the client assess whether AI techniques could better categorise risk events. We designed a machine learning model using R as the underlying programming language.
Crowe applied language analysis to risk event descriptions and used unsupervised learning to let the algorithm categorise events independently. We also mapped patterns from a labelled dataset to the categories identified through unsupervised learning.
This leading insurance-focused operational risk member organisation was looking to enhance the efficiency and effectiveness of its insurance operational risk capabilities through the anonymised and confidential exchange of data about risk events between firms.
Approach
Challenge
Design and deployment of a bespoke machine learning model to improve operational risk event categorisation.
The client wanted to explore the extent to which AI tools, including machine learning, could help categorise risk events to drive efficiency and consistency, reduce time and enable better insights.
Case study
How we have helped our clients
Outcome and benefits