Insurance customer service enhancements with Amazon cloud

Understanding the challenge

Our client, a Fortune 100 multi-line insurance provider specializing in life, annuities, retirement, and worksite products, faced challenges in delivering customer service through direct mail and live calls. This labor-intensive process resulted in long hold times and a negative customer experience. To address these issues, the client began using the Amazon cloud platform but had limited technical maturity. They aimed to enhance their customer experience, increase the organization’s technical capabilities, and accelerate the design, development, and deployment of an omnichannel self-service experience.

The PwC solution

To address the client's challenge of introducing self-service capabilities across voice and digital channels, PwC implemented a strategic approach by developing an end-to-end delivery model. The objective was to identify the primary contact drivers, understand customer personas, and determine customer intents early in the process. This enabled effective routing of requests to the appropriate skill destinations, establishing a seamless omnichannel experience.

• Identifying contact drivers: We focused on identifying the top customer intents and helped implement a Natural Language Processing (NLP) Virtual Assistant to enhance the overall customer experience and increase containment rates across various business units.
• Cross-functional team: We assembled a cross-functional team and established a governance forum with clear roles and responsibilities. This facilitated consensus in decision-making, cataloging processes, transparency, and engagement across teams. Tactical decisions, such as conversation design, were made with ultimate business outcomes (e.g., cost savings and business continuity) in mind, supported by holistic reporting on operational and business value metrics.
• Conversational experience design: Our team of linguists and conversational experience designers meticulously crafted customer personas, infusing empathy and a human touch into each interaction. This strategic approach effectively prevented customers from falling into repetitive loops. Drawing from a robust pool of data sourced from diverse customer touchpoints, including demographics, behavior patterns, purchase history, and feedback, we embarked on a thorough segmentation analysis. This enabled us to categorize customers into distinct groups with similar characteristics and needs, ranging from new customers to frequent users, from those with specific product interests to those requiring frequent support. Subsequently, for each segment identified, we embarked on a detailed journey mapping exercise. This involved tracing the customer journey from the initial point of contact to resolution, meticulously pinpointing key interaction points and pain points along the way. Through this process, we aimed to enhance satisfaction and operational efficiency by addressing critical touchpoints. In tandem with journey mapping, we delved into persona development for each customer segment. These detailed personas served as guiding beacons, shaping the design of personalized interactions and support services tailored to the unique needs and preferences of each customer group.
• AI development: We structured Lex Bots logically, enabling them to handle intents and sub-intents effectively without needing fallback mechanisms. This allowed for a seamless and holistic customer experience.
• AI testing: We helped implement an iterative testing, training, and tuning cycle to continuously improve Natural Language Understanding (NLU) detection. As more intents were released, we anticipated potential impacts on NLU accuracy. To maintain targeted containment rates, we introduced iterative testing, training, and tuning processes, enabling consistent customer experiences across channels.
• A/B testing and deployment: To avoid regression post-production deployment, we employed A/B testing strategies across channels. We utilized centralized AI to power all customer contact channels, load balancing, and diverting a small percentage of traffic to newly developed NLU systems. Traffic to the new NLU was gradually increased in a controlled manner while continuously evaluating performance. This holistic approach allowed for a seamless, efficient, and empathetic customer service experience across channels.

Deeper dive into AWS technology used

The client's AWS-hosted solution tapped into Amazon Connect Cloud Contact Center, offering customers multiple channels for engagement. To swiftly grasp customer intent, we employed Amazon Lex, trained with custom client dictionaries and three years' worth of voice transcript data from Amazon Transcribe. For seamless omni-channel experiences, spanning phone calls, web chats, and SMS interactions, we utilized DynamoDB to construct a high-performance, low-latency unified customer data platform. Challenges like intent fallback led us to create a bot orchestration layer using AWS Lambda and Lex APIs. Additionally, we helped implement an Amazon Connect Outbound campaign paired with Amazon DynamoDB for callback scheduling, prioritizing human-to-human connections. Our team continually updated the web chatbot front end and NLUs, leveraging AWS Code Commit, Code Build, Code Deploy, and CodePipeline for efficient release management. For monitoring and alerting, we relied on AWS CloudWatch logs, Metrics, CloudWatch Events/CloudWatch EventBridge, and Alerts, confirming prompt notification of stakeholders. We integrated a next-gen NLP-based search service to navigate organization-wide knowledge bases, SharePoint, and cloud storage using S3 and Amazon SageMaker. To meet the client's requirement for a customized agent workspace, we developed a single-page application in Angular.JS and utilized Connect Streaming APIs for desktop customization. Lastly, for building a highly available contact center, we recommended starting with a Single Region approach and replicating critical applications using Amazon Connect Global resiliency (ACGR), meeting RPO and RTO objectives while optimizing costs.

Delivering outcomes

The outcome showcases a fully operational Omni channel Cloud Contact Center, seamlessly integrating Voice, Chat, and SMS channels, complemented by a Customized Agent Desktop and advanced Conversational Analytics utilizing Contact Lens. The implementation of AWS Connect and Lex not only supported the omnichannel future state call center but also introduced a critical feature development known as the intent engine. Furthermore, in addition to these capabilities, over 20 core Conversation as a Service (CaaS) AWS Lex Chatbot functionalities were developed. These encompassed NLP, AI/ML, and Voice transcription technologies, enhancing the customer self-service experience while driving operational efficiency. Additionally, a next-gen NLP-based search service was deployed to access organizational-wide knowledge bases, further enhancing customer interactions. The integration of AWS Connect facilitated critical features such as live chat, queue routing, and request call back, resulting in a remarkable 70% call deflection from traditional call centers. Moreover, there was a notable 60% increase in agent daily call handling capacity across all channels, with the new platform continuously learning and improving through active-learning mechanisms. To bolster performance evaluation, robust reporting capabilities were developed, enabling insights into CSAT, AHT, and other experience drivers, crucial for identifying key factors contributing to poor experiences. Finally, the solution empowered real-time monitoring of conversations by contact center managers, granting them the flexibility to intervene in live interactions, thus enhancing the overall customer experience.

The results yielded by the implementation of Amazon Connect Contact Center Delivery stand poised to unlock a myriad of fresh business prospects and catalyze revenue expansion for our esteemed client. Our efforts extended beyond mere deployment; we embarked on a journey to help equip the client with proficiency in industry leading AWS technologies, thereby enabling them to chart a course towards self-sufficiency.

Leveraging AWS technology, PwC's solution helped to seamlessly integrate AI across customer contact channels and automated complex legacy leave absence management processes, catapulting the client to the forefront of their industry. This implementation yielded substantial time and cost savings, driving improved operational efficiency and enabling consistency across channels with a remarkable containment rate of over 70% and a 60% boost in agent productivity. These outcomes have not only improved the client's business processes but have also positioned them for heightened sponsor satisfaction, new avenues for business expansion, and sustainable revenue growth.

Explore how PwC and AWS are accelerating breakthrough outcomes.