Case Notes/Answers
Concentrix Corporation Improving Customer
Persistency for an Indian Insurance Company By Shylu
John, Bhavin Shah
Discussion Questions:
1. Identify the possible options for improving customer persistency at a minimal cost and discuss the
strengths and weaknesses of each option. Assume that the cost per call is ₹30.00, cost per SMS is
₹0.15, and cost per email is ₹0.00.1
2. What are the various predictive modelling options available to Akhilesh Kumar, the member of
Khanna’s interdisciplinary team who built the predictive model? Why does logistic regression fit
here?
3. Using the Student Spreadsheet (Product no. W25823), suggest the definition of the target variable
to be used for the payment before due date (PDD) model that estimates the percentage of policies paid
on or before due date and past due date.
4. Develop models with and without historical payment behaviour (Mean_NDPD) as a variable and
advise management of the key findings.
5. Describe in detail the steps followed to build the models and the key issues or challenges faced
during model building.
6. Compare the matrix- and model-only approaches with the validation data set and recommend the
best choice for the contact strategy.
, W25822
Teaching Note
CONCENTRIX CORPORATION: IMPROVING CUSTOMER
PERSISTENCY FOR AN INDIAN INSURANCE COMPANY
SYNOPSIS
Concentrix Corporation (CNX) partnered with Photon Life Insurance Company (PLI), a leading insurance
provider in India, in August 2017 to support PLI’s customer management service. To serve PLI, Mohit
Khanna, global operations manager at CNX, set up a customer management operations team in India with
forty agents for outbound calling. On February 1, 2018, he undertook a promising task to improve customer
persistency at minimal operational costs. To understand customer persistency, policies were tracked at
intervals of the thirteenth, twenty-fifth, thirty-seventh, forty-ninth, and sixty-first month post-policy
enrollment. PLI had a low persistency ratio of 14 per cent at the sixty-first month and was ranked fifteenth
in the industry based on this metric in 2015–16.
The process at that time was for available agents to call or send short messaging service (SMS) text
messages to randomly selected policyholders. This was found to be highly ineffective, with limited to
negligible improvement in persistency. If all policyholders with an approaching due date were to be called,
then the number of service agents required would be very high, leading to significant operational cost
increases. Khanna’s challenge was to design a contact prioritization strategy that could be implemented
with minimal operational cost and result in improved persistency.
LEARNING OBJECTIVES
Working through this case offers students the opportunity to do the following:
• Understand customer persistency in the insurance industry and its management by contact centre operations.
• Understand predictive modelling approaches and the selection of best fit model.
• Use of a predictive analytics solution in a contact centre strategy to solve the business problem.
• Formulate a cost-effective contact centre strategy to improve customer persistency.
This Teaching Note is authorized for use only by ELENA PIKULINA, University of British Columbia until Feb 2025. Copying or posting is an infringement of copyright.
or 617.783.7860.
Concentrix Corporation Improving Customer
Persistency for an Indian Insurance Company By Shylu
John, Bhavin Shah
Discussion Questions:
1. Identify the possible options for improving customer persistency at a minimal cost and discuss the
strengths and weaknesses of each option. Assume that the cost per call is ₹30.00, cost per SMS is
₹0.15, and cost per email is ₹0.00.1
2. What are the various predictive modelling options available to Akhilesh Kumar, the member of
Khanna’s interdisciplinary team who built the predictive model? Why does logistic regression fit
here?
3. Using the Student Spreadsheet (Product no. W25823), suggest the definition of the target variable
to be used for the payment before due date (PDD) model that estimates the percentage of policies paid
on or before due date and past due date.
4. Develop models with and without historical payment behaviour (Mean_NDPD) as a variable and
advise management of the key findings.
5. Describe in detail the steps followed to build the models and the key issues or challenges faced
during model building.
6. Compare the matrix- and model-only approaches with the validation data set and recommend the
best choice for the contact strategy.
, W25822
Teaching Note
CONCENTRIX CORPORATION: IMPROVING CUSTOMER
PERSISTENCY FOR AN INDIAN INSURANCE COMPANY
SYNOPSIS
Concentrix Corporation (CNX) partnered with Photon Life Insurance Company (PLI), a leading insurance
provider in India, in August 2017 to support PLI’s customer management service. To serve PLI, Mohit
Khanna, global operations manager at CNX, set up a customer management operations team in India with
forty agents for outbound calling. On February 1, 2018, he undertook a promising task to improve customer
persistency at minimal operational costs. To understand customer persistency, policies were tracked at
intervals of the thirteenth, twenty-fifth, thirty-seventh, forty-ninth, and sixty-first month post-policy
enrollment. PLI had a low persistency ratio of 14 per cent at the sixty-first month and was ranked fifteenth
in the industry based on this metric in 2015–16.
The process at that time was for available agents to call or send short messaging service (SMS) text
messages to randomly selected policyholders. This was found to be highly ineffective, with limited to
negligible improvement in persistency. If all policyholders with an approaching due date were to be called,
then the number of service agents required would be very high, leading to significant operational cost
increases. Khanna’s challenge was to design a contact prioritization strategy that could be implemented
with minimal operational cost and result in improved persistency.
LEARNING OBJECTIVES
Working through this case offers students the opportunity to do the following:
• Understand customer persistency in the insurance industry and its management by contact centre operations.
• Understand predictive modelling approaches and the selection of best fit model.
• Use of a predictive analytics solution in a contact centre strategy to solve the business problem.
• Formulate a cost-effective contact centre strategy to improve customer persistency.
This Teaching Note is authorized for use only by ELENA PIKULINA, University of British Columbia until Feb 2025. Copying or posting is an infringement of copyright.
or 617.783.7860.
