The Auto & Home Insurance group ("Client") of a leading diversified US based Financial Advisory company was using data from multiple sources (Internal sources, Credit Bureaus and Business Partners) to develop its marketing strategies and business decisioning. More specifically, the client was deploying predictive analytics techniques on data from these sources for the following:

• Increasing effectiveness of Marketing Campaigns conducted across various channel partners (Retail, Airline, Auto Manufacturer, Travel Advisory etc)
• Reducing Customer Acquisition costs
• Identifying lucrative and/or risky customer segments for Auto & Home Insurance products
• Developing decisioning logic for evaluating Motor Vehicle Records (MVR) and Loss Underwriting.

The analytics team responsible for executing these tasks was "short staffed". The client was exploring mechanisms of scaling up their internal analytics team quickly in a cost effective manner to meet demands of their business

Inductis proposed to augment the client's internal analytics team by deploying an experienced cross-functional team of Marketing & Risk Analytics professionals to provide "high quality" analytics support through a cost effective offshore based delivery platform.

The team was quickly assembled from Inductis' large pool of experienced India-based analytics resources. We deployed a cross functional team of Programmer Analysts and Analytics Team Lead to bring in multi-pronged expertise in SAS programming, statistical knowledge and business acumen tailored to this client's needs.

The team was structured in the following manner to enable seamless working and communication at different levels for defining project/activities, executing initiatives, monitoring progress and discussing results:

Team structure click here

As a first step (Transition Phase), we worked closely with the client to lay out a transition plan for offshoring the analytics work. The plan included details around:

1. Roles and Responsibilities of stakeholders
2. Resource Management Guidelines and Reporting Requirements
3. Activity Plan for Team
4. Any Training Requirements for a specific need
5. Onsite Travel Requirements / Plan for the entire engagement
6. Communication Protocols
7. Project Management Protocols
8. Data Transfer Mechanisms
9. Necessary Technology Infrastructure Requirement
10. Success Metrics and Methods of Evaluation

In the next Phase, the Inductis team started executing the work according to the transition plan. The team followed detailed communication and data/knowledge transfer protocols to ensure seamless functioning like an extended internal client team. The Inductis Analytics Team Lead worked closely with the client to ensure the following:

1. Sufficient Project pipeline was built up for the team
2. Issues were resolved through daily, weekly and monthly updates
3. Project deliverables, quality and deadlines were strictly adhered to by the team
4. Productivity, resource utilization and variance from budget was tracked and communicated
5. Knowledge management material was created and shared with the client as and when needed
6. Project feedback and performance rating was captured from Client

The team worked on the following over the course of 12 months:

• Building Response and Conversion Predictive models for increasing the effectiveness of marketing campaigns
• Assessing alternate data sources for reducing cost of purchasing expensive credit bureau and MVR data
• Segmenting customers based on demographics and risk profiles for designing customized mail campaigns
• Building predictive models specific to auto-insurance related behavior (For example, predicting a bad Motor Vehicle Record)
• Building Look-alike models for identifying lucrative customers
• Profiling risk categories of the customer portfolio

The Inductis team used the SAS9 Platform in conjunction with several non-parametric tools like Classification and Regression Trees (CART) and Multivariate Adaptive Regression Splines (MARS) to execute these projects. We deployed our knowledge of the insurance industry to design optimal solutions for our client within the constraints of available data.

We followed a blend of our proprietary methodology coupled with clients' methodology for creating predictive models, thus enabling a fruitful cross-pollination of advanced techniques for delivering high quality solutions.

We evaluated model performance on statistical metrics that included Multicollinearity checks, % Concordance, Area under ROC, KS Statistics, Hosmer-Lemeshow test etc. Further, we tested the robustness of the model by analyzing variations in model parameter coefficients through techniques like Coefficient Blasting & Boot Strapping.

We provided our client with detailed documentation to facilitate their implementation of the models in their environment.

Significant Costs Savings
Our team helped the client gain ~25%+ cost savings in their direct resource costs (salary and bonus) while meeting their internal hiring goals. Further, the hidden costs associated with recruitment, training, skill set development, attrition management etc. were taken out of clients' system as we invested in it at our end.

High Quality Output
The cross-functional structure of the team allowed it to work comfortably across a breadth of business problems of varying degree of complexity. The high rigor of analysis conducted by the Inductis team generated several new insights into the business. Further, recalibration of existing and new models provided consistent model performance over different quarters.

Successful Offshore Delivery Model
Our team outperformed the expectations of the client on all parameters, including smooth delivery of projects from the virtual offshore team and quick ramp up of additional resources on a short notice. The client was able to successfully integrate the analytics support provided by Inductis in marketing & customer acquisition efforts.

Inductis team gained these efficiencies by:

1. Efficiently hosting approximately 150+ GB data from different sources in an integrated manner allowing the team to readily perform any analysis faster

The significant impact of deploying an offshore Analytics team from Inductis is clearly obvious from the above case study. We, at Inductis have provided this impact to several other clients through similar constructs. If you are interested in discussing a similar arrangement, feel free to write to asaluja@inductis.com