AWS Public Sector Blog

Improving customer experience for benefits eligibility and enrollment

laptop next to a pile of paperwork and a pen

State governments are currently experiencing a spike in the number of applications for social benefit programs. These programs serve millions of people every year to support healthcare and to help keep them out of poverty. To address the surge, agencies are looking at cloud-based services including artificial intelligence (AI), conversational bots, and call centers to improve the customer experience as well as workforce productivity, provide more visibility for agency leadership into program performance, and help reduce fraud, waste, and abuse.

Current challenges for agencies and beneficiaries

In fiscal year 2019, the United States spent $2.5 trillion on social benefits programs serving over 100 million citizens. To help with benefits enrollment, state agencies use integrated eligibility and enrollment systems (IE). To receive benefits, constituents go through complex application, claims, eligibility, enrollment, and adjudication processes. In most cases, beneficiaries wait several weeks before their benefits are approved due to the high volumes and complex workflows involved in processing these applications. Case managers and workers spend a significant amount of time with manual reviews and deal with legacy systems, resulting in large application backlogs and a poor customer experience.

The recently passed Coronavirus Aid, Relief, and Economic Security (CARES) Act extends benefits to new categories of workers, expanding the pool of people eligible to apply for unemployment insurance (UI). While UI systems are the first to feel the impact of unemployment, health and human services (HHS) programs that help people who are unemployed—including Supplemental Nutrition Assistance Program (SNAP) and Medicaid —also experience growth in the volume of applications. Agencies also deal with fraud, waste, and abuse that have wide impacts on the delivery of benefits. The recent whitepaper, “Augmented Artificial Intelligence (AI) – The power of human + machine” discusses these challenges in detail and proposes solutions to address them.

How the cloud can help improve benefits delivery

Cloud capabilities from data lakes to machine learning (ML) to Amazon Augmented AI (A2I) can help improve delivery of benefits by:

  • Providing high visibility into program operations: Data-driven insights enable agencies to build programs and advocate for innovative policy changes to better serve constituents. Benefits administrators can enable program leadership to make decisions by using data lakes, data analytics, and ML. Data lakes are supported by AWS services such as Amazon Simple Storage Service (Amazon S3), AWS Lake Formation, and AWS Glue. Both real-time and batch analytics provide insights into program operations metrics like backlogs and processing times. Agencies can improve program preparedness by using AI and ML for forecasting budgets and enrollment models and use ML models to detect and prevent fraud, waste, and abuse, through offerings such as Amazon SageMaker, Amazon Forecast, and Amazon Fraud Detector. As an example, the Financial Industry Regulatory Authority (FINRA) uses AWS to power their big data analytics pipeline that handles 135 billion events per day to help monitor the market, prevent financial fraud, protect investors, and maintain market integrity.
  • Modernizing benefits systems: Legacy benefits systems are often inflexible and cannot scale or accommodate agile development. Using modular and microservices-based architectures, containers, and/or serverless technologies, agencies can build systems that are flexible to handle program or policy changes and enable advanced analytics, service delivery improvements, and digital transformation. HHS agencies including Maryland DHS and Center for Medicare and Medicaid Systems used AWS to modernize their mission-critical applications that administer healthcare and social benefits programs for millions of beneficiaries.
  • Modernizing contact centers: Agencies can quickly deploy a cloud-based contact center using Amazon Connect and allow agents to work anywhere. Because Amazon Connect operates on a pay-as-you-go basis, as call center volume decreases over time, so does the cost of the call center. For example, the Kansas Department of Labor received as many as 1.6 million calls from residents needing guidance on unemployment claims. They’re using AWS to sort and direct to the correct representative. Also, LA County achieved 60 percent cost savings and a 17 percent reduction in call volume using Amazon Connect. In addition to the cost and scalability benefits, agencies can use ML to get insights on contact center operations like customer sentiments and call transcript analysis to make decisions on staffing and training.
  • Improving beneficiary/customer experience: Providing digital capabilities for an end-to-end application process including submission, status, interview, and approval can enable agencies with self-service and automated communications via web, mobile, and contact centers using cloud offerings such as Amazon Pinpoint, Amazon Lex, and Amazon Polly. Using chatbots developed with Lex, beneficiaries can get their case status, execute routine tasks (such as a PIN resets), or obtain general information on claims. Users can also submit mobile documents, schedule interviews online, verify financials, and more. AWS Amplify helps build secure and scalable mobile and web applications to enable some of these features. For example, using Amazon Polly, West Virginia was able to handle 96 percent of their UI-related calls using an interactive voice response (IVR) and only four percent required a live agent.
  • Improving workforce productivity: Automating the eligibility review and approval process to minimize manual tasks can improve workforce productivity and allow the workforce to focus on more complex cases. Amazon Textract, AI, and ML capabilities can be used to flag and report data anomalies speeding up reviews and approvals. A2I provides built-in human review workflows for common ML use cases, such as text extraction from documents, which allows predictions from Amazon Textract to be reviewed. A2I makes it simple to integrate human judgement and AI into virtually any ML application.

Getting started

Download the “Augmented Artificial Intelligence (AI) – The power of human + machine” whitepaper to learn more including a reference architecture and a framework to improve the benefits service delivery for citizens and contact us at aws-hhs@amazon.com or reach out to your AWS Partner Network (APN) Partner to get started.