Overview

Vantwell Insurance is a US based insurer offering auto, home, and life coverage. It held years of policyholder data, but the data sat in separate policy, claims, billing, and support systems, so the same customer showed up as several disconnected records and no team could see the full relationship. Retention was reactive, teams learned a policyholder was leaving only once the policy lapsed, cross-line selling ran on guesswork, and every business question turned into a slow, backward looking report. Bacancy built a unified policyholder profile, then applied customer data analytics and AI to predict lapse risk before renewal, rank cross-sell propensity, and segment policyholders by value. The insights reached business teams through governed dashboards they could use directly, with masking and role based access on regulated data and explainable model outputs for compliance. Vantwell moved from reactive reporting to predictive engagement, catching at risk policyholders earlier, targeting cross-sell with far less waste, and answering routine questions the same day.

Technologies Used

Snowflake
Python
dbt
Scikit-learn
Power BI Embedded
Git

Project Highlights

checkmark

Fragmented policyholder records unified into a single customer profile per person

checkmark

Customer data analytics and AI applied to lapse risk, cross-sell propensity, and value segments across auto, home, and life

checkmark

Regulated policyholder data protected with masking and role based access throughout

checkmark

Predictive insights delivered to business teams through governed self-serve dashboards

The Challenges

1

Policyholder data was split across policy, claims, billing, and support systems and across product lines, so the same customer existed as several disconnected records and no team could see the full relationship.

2

Retention was reactive. Vantwell learned a policyholder was leaving only when the policy lapsed or a cancellation call came in, with no early signal to act before renewal.

3

Cross-sell ran on guesswork. With data siloed by product line, a customer who held one policy but fit others well stayed invisible, so campaigns went out broadly and converted poorly.

4

The data included regulated PII that could not be exposed during analysis, and any model that shaped how a customer was treated had to be explainable and checked for bias to satisfy compliance.

Solutions by Bacancy

1

Bacancy’s data analytics services team built identity resolution to match and merge records across systems and product lines, giving Vantwell one profile per policyholder and a single view of the whole relationship.

2

Our data science and predictive analytics team built machine learning models that scored each policyholder’s lapse risk ahead of renewal, so retention teams could act early on the accounts most likely to leave.

3

We ranked cross-line propensity for every policyholder and group them into value based segments, so marketing could target the right insurance customer with the right offer instead of broad, low converting blasts across every policy line.

4

Our insurance IT services covered access control on the analytics layer, using column level masking and role rules so each user saw only their allowed policyholder and claims data. Model outputs stayed explainable, so decisions held up in compliance review.

Core Features

checkmark

Identity resolution for a single policyholder profile across products

checkmark

AI lapse and churn prediction scored ahead of renewal

checkmark

Cross-sell propensity ranking and value based segmentation

checkmark

Governed self serve dashboards with masking and explainable model outputs

No. of Resource

04

No. of Resource

Time Frame

September 2025 – March 2026

Time Frame

Project Snapshot

Dashboard

Outcomes

At risk policyholders flagged 60 to 90 days before renewal, instead of at the cancellation call

Roughly 18% lower lapse rate among the targeted at risk segment after early intervention

Cross-sell conversion up around 22% from propensity targeted campaigns versus broad campaigns

Three to four fragmented records per customer collapsed into one policyholder profile

Routine business questions answered same day in dashboards, down from multi day report requests

All policyholder PII masked with role based access, and model outputs explainable for compliance review

Technical Stack

Cloud data warehouse Snowflake
Data ingestion Fivetran
Data transformation DBT
Orchestration Apache Airflow
Identity resolution Python
Machine learning Python, scikit-learn, XGBoost
Model tracking MLflow
BI / dashboards Power BI
Security & governance Snowflake RBAC, dynamic data masking
Version control & CI/CD Git, GitHub Actions

Experience With Bacancy

2500+ Projects Experienced Innovation with Bacancy!

Get access to an experienced team of developers and engineers from Bacancy, handpicked to ace your goals. Kickstart within 48 hours, no-risk trial.

Book a 30 min call

14+

Years of Business Experience

1458+

Happy Customers

12+

Countries with Happy Customers

1050+

Agile Enabled Employees

How Can We Help?