Specialty Insurance Carrier

Identifying and delivering three new data science models with $20 million annual EBITDA potential

How do you find the best opportunities to use data to create financial value?

Our client, a specialty insurance carrier, wanted to tap into its vast data to capture competitive advantage and improve financial performance. We used our proprietary data monetization approach to deliver three new data science models with the potential for adding $20 million to EBITDA annually.

200%

Profitability increase by decreasing average claim severity through ML-recommended inspections

$3M

annual risk avoidance through improved dealer onboarding

$10M

in new revenue by identifying and targeting new markets

The Challenge

To capture advantage in the competitive middle-market insurance sector, companies need to leverage every tool at their disposal. But there’s one that many still struggle to exploit: data.

Our client wasn’t taking full advantage of its considerable volume of data. Executives knew there were opportunities to use data to run more efficiently and compete more effectively—resulting in direct EBITDA impact. That’s where we came in, with data at the center of our ability to create financial value.

Project Timeline

weeks

Ideate and quantify potential data science use cases for two business units

weeks

Develop three prototypes for sign-off

months

Build and refine the data science models

weeks

Test and take the models live

An Undeniably Different Approach

After previously implementing Databricks and Azure Container Apps, we had the foundation for a data science and MLOps program to achieve value in future use cases. Our multidisciplinary team consisted of insurance, product, commercial analytics, and data engineering/visualization experts who employed our proprietary data monetization framework.

We began with the big picture—generating hypotheses about data products for two business units. Our team used value identification analytics principles to test hypotheses with the client’s data and shared the potential financial impact with them. With the client’s confidence in the value creation story, we then built and implemented the models, integrating them seamlessly into the client’s technologies and processes.

Our client now has three customized predictive models that put insight in the hands of decision-makers at critical points—enabling them to drive performance, profitability, and advantage:

Claims adjudication scoring model that delivers guidance to adjudicators using machine learning for approving, rejecting, or further investigating individual claims in just seconds.
New dealer profitability model that helps business stakeholders determine which new dealers to onboard, and with which products based on profitability projections.
Segmentation/underwriting model that generates insight for strategically targeting risk categories.

After implementation, we measured impact to ensure the models were creating the value expected. We set our client up for long term success by automating dashboard for ongoing monitoring and standing up a data organization to maintain them and create new models.

Real Results

Once live, the three models delivered results translating to $20 million in potential annual EBITDA gains—through a combination of cost reduction, risk avoidance, and increased revenue. Looking closer:

The claims adjudication model reduced costs, potentially adding $7 million to EBITDA annually—a 200% increase.
The new dealer onboarding model mitigated risks of onboarding unprofitable dealers—potentially adding $3 million annually to EBITDA.
The segmentation analysis model added $500,000 in profit through a program with $10 million in annual revenue—and that’s just a start, as this flexible model can support other parts of the business.

This project didn’t just accelerate our client’s analytics and data science program; the initial results opened eyes in the organization about the possibilities for creating financial value.