AWS Public Sector Blog

Reimagining unemployment contact centers

AWS branded background design with text overlay that says "Reimagining unemployment contact centers"

Benefit programs like unemployment insurance are complicated and time-sensitive with weekly eligibility requirements. As a result, contact centers routinely experience high call volume that increases exponentially during economic downturns. Yet for many constituents, contact centers are the only lifeline to get the answers they need while looking for work. More than ever, agencies can benefit from AI and automation to reduce call volume, improve self-service, and empower staff productivity—not just to maintain service levels during low unemployment, but to scale and rise to the occasion of unexpected surges in demand.

As a former state ombudsman, part of my job was to assist claimants with barriers and identify gaps in the customer experience. Why were customers calling? Where did the process break down for them? What populations were being underserved? Constituents with barriers are often disproportionately impacted when the standard channels are limited and overwhelmed. For example, callers without access to web applications or interactive voice response (IVR) systems in their native language are more reliant on connecting with a human agent and subject to prolonged wait times. Agent interactions conducted through an interpreter also tend to have longer average handle time.

So, how can agencies create capacity in their contact centers while improving equitable access to critical services? How do you scale and automate some of the more routine but time-consuming interactions to free up highly trained staff for higher value activities?

Amazon Connect is an omnichannel cloud-based contact center service born out of the real-world customer service needs of Amazon.com. With a full suite of features powered by AI and machine learning (ML), agencies can build natural, dynamic, and personalized self-service experiences. For example, constituents can interact with conversational AI for FAQs and receive proactive text or email updates in a state’s top languages, as opposed to navigating complex IVR menus and waiting in long queues. Integrated chat assistants can also provide real-time claim information through the IVR without agent assistance.

These solutions don’t just alleviate front line staff. They improve equitable access, opening new pathways to self-service for limited English proficiency (LEP) and low literacy populations, as well as individuals without access to computers or broadband. Every constituent has a different set of circumstances that dictates how and when they prefer to communicate. Some may lack a stable mailing address and prefer to interact online; others may be more responsive to text message than email. Amazon Connect provides the flexibility to reach customers where they are and improve their chances of compliance and success. Agents, supervisors, and administrators also have access to an array of AI-driven tools to improve efficiency, accuracy, and better understand customer pain points.

Figure 1. Demo of Amazon Connect for unemployment insurance. Take a journey through the first few weeks of an unemployment claim and explore how Amazon Connect can transform the contact center experience. (Demo by Greg Smith)

Strategies for reducing call volume and improving self-service

Over a dozen states have implemented Amazon Connect to support their unemployment contact centers. Although every agency is unique in its operational needs, in the next section, we map some of the capabilities of Amazon Connect to a four-pronged strategy for reducing call volume and improving self-service. We also share how states have already benefited from some of these approaches and best practices.

Figure 2. Chart aligning Amazon Connect capabilities to a comprehensive strategy for reducing call volume and improving self-service.

Decrease the number of reasons to call

Expanding channels to include voice, text, email, live chat, and chat assistant in a state’s top languages distributes call volume across multiple access points while improving the customer’s ability to self-serve in their native language. Moreover, generative AI now enables more conversational and multilingual experiences across these channels to unpack static content found in claimant handbooks or FAQs. With Amazon Lex and Amazon Bedrock, agencies can customize leading foundation models (FMs) with their unique help content to securely deploy sophisticated voice- and text-based chat assistants.

Agencies can also preempt potential calls with outbound campaigns, reaching up to millions of customers across channels with program updates and reminders for important deadlines, weekly certifications, and work search requirements. Event based outbound notifications can be personalized with real-time claim updates and timeframes to guide customers through the stages of a claim.

The Rhode Island Department of Labor and Training, for example, implemented chat assistants to help constituents navigate the claim process, and a mobile-friendly claim status portal that provided 24/7 access to claim information. The integrated platform sent automatic updates by text or email regarding a status change or a call to action, deflecting call volume and increasing the likelihood of customer compliance and timeliness.

Establish scalable IVR self-service options

Constituents view the contact center and web applications as a unified experience. If they can’t get through to an agent or have to navigate complex IVR menus for even the most basic answers, the entire experience suffers.

As a first line of service, multilingual voice chatbots provide callers with answers to general questions regardless of agent availability. Iowa Workforce Development relied on voice chat assistants to answer FAQs when claim volume exploded by more than 2,100 percent. The chatbot handled routine but time-consuming questions like, “What is unemployment insurance? How do I apply? How is my benefit rate calculated?” As a result, the agency was able to shift 30 percent of calls to self-service, saving close to 11,000 staff hours in a 3-month period.

API integration with backend systems also unlocks 360-degree customer experiences. Chat assistants can deliver real-time, personalized information through the IVR for routine inquiries such as claim status, weekly payment status, remaining weeks, overpayment balances, and benefit-year end-date. In Solving the Claim Status Question through Automation, North Carolina’s Division of Employment Security reported that 30 percent of incoming calls during the pandemic were related to claim status. To automate and scale responses to this top call driver, the agency implemented an IVR claim status module that deflected 20 percent of call volume when the agency was receiving around 200,000 calls per day.

Although not all customers will be able to self-serve through these interfaces, achieving a critical mass of adoption drives meaningful deflection for contact center managers, creating capacity for callers who really need the human assistance. Amazon Connect also includes monitoring tools to measure the effectiveness of these automation services and fine-tune them over time.

Improve load management

Voice- and text-based chat assistants maintain continuity of service even when the contact center is closed. They operate 24/7 to offset traffic and assist individuals who may have been unable to call during business hours due to work or other obligations. Meanwhile, agencies have the option to offer not just automatic callbacks, but caller scheduled callbacks so customers can select a callback window that is convenient for them.

The Connecticut Department of Labor achieved a 60 percent reduction in repeat calls after implementing IVR scheduled callbacks. Callers would receive a text message confirming their appointment time along with the option to cancel, keeping agent queues optimized. Additionally, customers could schedule calls online and indicate intent from a dropdown menu, which captured valuable data on how to route inquiries to the right skills. With these insights, the team was able to adapt staffing levels and generate proactive web content to address emerging customer questions.

Amazon Connect also includes vital workforce management capabilities to keep contact centers sufficiently staffed at all times. ML-powered features such as forecasting, capacity planning, and scheduling help agencies anticipate contact volume and arrival rates, convert forecasts into projected staffing needs, and assign daily shifts to the right number of agents (see demo).

Reduce average handle time

Every minute counts when you’re understaffed and call volumes escalate. Amazon Connect incorporates tools to boost agent productivity and measurably improve the customer experience.

First, Amazon Connect Voice ID streamlines the caller authentication process and enhances fraud risk detection with ML. Rather than requiring callers to share personal data elements over the phone, Voice ID analyzes a caller’s unique voice characteristics to provide agents and self-service IVR modules with a real-time decision on the caller’s identity. Voice ID also screens for fraudulent actors in real time, based on your contact center’s custom watchlist, reducing potential losses from fraudulent attacks.

Amazon Q in Connect is a generative AI assistant embedded in the agent workspace. Using conversational analytics and natural language processing (NLP), it interprets voice and chat interactions in real time, generating recommended responses and actions derived from an agency’s unique knowledge base content such as FAQs, websites, claimant handbooks, and training manuals. Agents can validate recommendations with links to source content and chat directly with the assistant for additional guidance. The assistant is highly useful for training new agents, scaling up contact centers, and supporting accuracy during ambiguous and evolving policy landscapes like the COVID-19 pandemic.

Lastly, Amazon Connect Contact Lens is a suite of AI-powered analytics and quality management capabilities that lets you monitor, measure, and continually improve contact quality and agent performance. Organizations gain valuable insights into customer sentiments, conversation themes, and emerging trends before customer experience issues become widespread (see demo). These insights can guide decisions such as updating IVR prompts, outbound campaigns, knowledge base content, and staff training. Contact Lens also uses generative AI to create post-contact summaries, reducing manual note-taking and after contact work (ACW) so agents can spend more time assisting customers.

Conclusion

Amazon Connect provides a broad range of capabilities that can be configured to execute a robust contact center strategy. Agencies can also add AI capabilities to their current contact center to activate self-service virtual agents, agent assist, and real-time and post-call analytics. Learn more by watching the demos in this post and explore case studies and related content below. To learn how Amazon Web Services (AWS) can support your unique needs, contact your AWS account executive, or complete this form. Also, visit the AWS for labor and workforce main page and the Do More with AWS page. AWS can support the journey with in-person and online trainings, acceleration programs, or professional services support.

Read related stories on the AWS Public Sector Blog

Related case studies and content

 

Contributing Author: Greg Smith, AWS cloud solutions architect