Association, Insurance

Business Impact Analysis and Advanced Algorithm POC Helps Motor Club Solve Membership Challenges

A proof of concept was created for a large motor club to identify trends present within the organization’s data and attempt to build a model to predict costly members and analyze potential changes to membership offerings.

LET'S BE BRIEF

Challenge

Ability to leverage data to identify current members that are high-risk and costly.

Objective

Analyze data using machine learning to identify motor club members misusing roadside assistance offering.

Solution

Proof-of-concept identifies patterns present within the organization’s data and is used to build a model to predict costly members and analyzes potential changes to membership offerings.

The proof of concept advanced algorithm increased the optics into existing members who are problematic and assigned them a risk rating for action to be taken. 

THE EXTENDED VERSION

Challenge

A large motor club recently revamped its data warehouse and the organization was seeking ways to maximize its data assets. The roughly 250 unique data points that were collected on every roadside assistance call presented a high volume of high-quality information to the organization. The motor club was challenged in its ability to easily leverage the data it had. More specifically, the organization had concerns regarding abuse and extreme cost within a subset of its customers and sought to analyze their data to identify the members misusing the roadside assistance offering.  

Solution

The engagement aimed to help answer important questions regarding members’ usage of the network such as:

  • Who are the current members who are most high-risk and costly?
  • As we acquire new members, who are the applicants most likely to be highly impactful?

A proof of concept advanced algorithm for identifying problematic users at a point in time was developed using additional features. This increased the optics into existing members who are problematic and assigned them a risk rating for action to be taken.

Additionally, a what-if analysis allowed the organization to see, at a glance, the financial impacts to multiple dimensions of their service offering to determine potential changes in their membership benefits.

How We Did It

Customer Lifecycle
Risk Identification & Abuse Detection
Trend Analysis

Tech Stack

  • Python
  • Postgres
  • Tableau
  • AWS