Insurance - Claim Process Automation via Email Data Extraction
As insurance businesses scale far and wide, their service desks are finding it an uphill task even to handle day-to-day claim-related email traffic and respond within SLAs. This builds a huge backlog to the point that it starts to hurt the performance and productivity of the service desk. Moreover, lack of timely responses and effective solutions to customer issues negatively impact a company’s credibility in the long run. A survey conducted by Super office reveals that the fastest response time from a service desk is 1 minute while the slowest is as long as 8 days, but on an average, the response time is nearly 12 hours. Above all, 62% of the companies do not respond to customer emails at all.
It is to relate to this reality when we realize what service desks face in terms of the extent of challenges such as:
Large, cumbersome email volumes that make customer support difficult
Tedious and redundant manual task of email data extraction
Complexity of tracking and recording service history
Time spent in understanding the context of emails
Constant pressure of having to provide quick responses
Our Intelligent Automation (IA)-based solution enables organizations to squarely address this challenge by bringing certain fresh and innovative techniques into the fray.
The solution is powered by artificial intelligence and leverages machine learning Technology (MLT) to respond to customer’s emails in real time. This solution can clearly assess the polarity of the email text and determine whether the customer tone is positive or negative or neutral by employing Natural Language Processing (NLP) to process the text. It comprises an internal machine learning engine with a threshold of prebuilt automated resolution and can effectively address 70% of the queries. For instance, when an email comes to the service desk, the IA solution parses the email content, matches the terms with its pre-built engines, and responds with a corresponding solution to the customer within a few moments. This application is integrated with core systems for gathering important customer-related information to put together an appropriate customized response. When the subject matter of the mail does not match with the pre-built engines for example, the system automatically raises a service ticket for the query and diverts the query to a subject matter expert for further processing.
The automated email solution can be categorized into three types:
Automated Complete Response: For every claim intimation email received, the MLT identifies and categorizes the email text, matches it with one of the ML Engine’s pre-built resolutions and sends an instant response to the Customer. (E.g. What is the claim procedure? What is the procedure for changing the payment mode?).
Automated Partial Response: If the claim/risk data in the email does not match with any of the pre-built resolutions, it is diverted to a customer representative for manual response while simultaneously getting back to the customer with the new service request details and an assurance that they will hear from the service desk with a final resolution.
Straight-Through-Processing (STP): In simple and straightforward cases, the system would automatically do a straight-through-processing and resolve the customer request. A claim request is created with a copy to the customer. (E.g., for a simple maturity claim where the benefit amount needs to be paid, the system would authorize the finance department to credit the amount to the customer’s registered bank account).
A significant feature of our solution based on Intelligent Automation is its self-learning feedback mechanism. The ML Engine constantly learns from the new entries and continues to enhance the automated response repository.
Customer sends an email to the company’s service desk
The service desk representative (SDR) reviews the email in the arrival sequence
If the content of the email is simple, the SDR responds quickly and closes the request
If the content of the email is complex, it is diverted to a specialist
The specialist responds with appropriate remarks in due course and closes the request
Revised IA-based Process
Customer sends an email to the company’s service desk
The service desk Intelligent Automation solution parses the email content
If the content of the email matches one of the prebuilt resolutions, an automated response is sent to the customer
The request is closed
If the content of the email requires further review, the service desk acknowledges the customer with a service request (SR) number advising the customer to quote the service number in future correspondence
The system diverts the query to a service representative for a detailed response
The service representative responds to the query in due course
The service request is finally closed Technologies
Tensor Flow, Python for Machine Learning
NLTK for Natural Language processing
Operational efficiency increased by 30-40% as simple requests such as password reset, procedures, product brochures, account statement, and premium paid certificate can be automatically process using Intelligent Automation
10% quarterly growth in the automated response repository due to self-learning feedback
Reduction by 20% in labour costs YoY
SME time saved can be utilized for other critical business