Applied Customer Engagement: Optimizing the Citizen Experience with Virtual Agents

Organizations everywhere are in the midst of a customer experience revolution — and the government is not exempt.

"There's this idea of liquid expectations: Citizens' expectations for the government are often set by the experiences they have with best-of-breed service providers. We see the speed, ease, and personalization provided by those interactions and, like liquid, those expectations flow into other parts of our lives, including our encounters with the government,” says Chris Zinner, managing director of Digital & Customer Experience for Accenture Federal Services.

While these liquid expectations have existed for many years, Zinner believes expectations are now proliferating, spurred by the recent pandemic.

“When you consider the impact of COVID-19, where face-to-face interactions are limited or impossible, the importance of having on-demand digital access to these services is becoming even more critical,” he notes.

As agencies face a massive surge in citizen inquiries, this increase in demand is hitting an already stretched system. Many of today’s agencies rely on older systems and limited staff, meaning workers at call centers and other customer service departments often have difficulty addressing customer needs quickly and efficiently. This situation often leads to longer wait times and lower customer satisfaction rates.

AI-backed tools and virtual assistants can help change that. By pairing digital tools with human staff, agencies can better manage increased contact volumes – via the phone and digitally – to consistently deliver effective customer service. Using these tools to offer options like 24/7/365 availability and more intelligent self-services means agencies can serve more people, provide more timely and personalized service and advice, and achieve higher customer satisfaction, often at a lower cost-to-serve.

Accenture developed its Applied Customer Engagement (ACE+) platform to deliver on this promise. ACE+ looks to revitalize contact center operations by tapping machine learning to analyze real-world interactions, uncover new insights, and equip virtual and human agents with the ability to deliver services more effectively.

“Digital – especially AI-driven customer care – has the opportunity to break the perceived tension that has existed for decades, which is the idea that to deliver a better experience, you have to invest more. And if you want to drive costs down, you have to reduce service levels and capabilities,” says Zinner. “Infusing AI into our digital tools and platforms can deliver more intelligent customer care with the opportunity to bend the cost curve as well.”

But what exactly do these digital tools look like — and how can agencies put them to work?

At the most basic level, agencies can implement simple, rules-based chatbots to respond to frequently asked questions (FAQs). Useful but not groundbreaking, as this information is often available online already, and most issues require further context to address fully. A cut above is a virtual assistant that uses natural language processing to host real-time conversations, allowing it to respond to a larger pool of more complicated requests by deciphering the user's intent.

These services offer the types of experiences that many have always wanted: virtual assistants work 24/7 and are typically available without a wait, providing a level of convenience and speed that human operations can't always reasonably accommodate. Moreover, these services can be delivered across multiple channels, text, apps, online portals, and more.

What ACE+ sees as the future of customer service is the intelligent agent, which combines customer understanding with reasoning capabilities to offer personalized responses to a customer’s specific question, problem, or need. These intelligent agents, according to Zinner, are what can enable agencies to truly transform self-service tools into more conversational and intelligent AI solutions.

"For someone to have confidence that a virtual agent can serve them adequately, we need to create a system using virtual agents that more closely mimics a human conversation and can understand multiple threads and understand the information in context," says Zinner. This is what intelligent agents can achieve, with the help of ACE+.

One of ACE+’s most valuable differentiators is human-centered design (HCD), which seeks to ensure that these intelligent virtual agents are architected in a way that increases empathy and understanding, connecting the pieces of the puzzle that comprise today’s customer experiences — across channels and vendors.

"We use human-centered design principles to immerse ourselves into the environment and context of those using these products — the client, customer service agent, and agency customer service leadership. This insight helps us make sure we're meeting everyone's needs and expectations across the ecosystem, and learning from every interaction along the way to continue to optimize the solution."

It’s this focus on HCD — alongside a vendor-agnostic approach that seeks to use leading vendor technologies from across the tech ecosystem — that allows Accenture to help agencies make the best use of conversational AI tools. In many cases, they rely on the Accenture Federal Digital Studio's capabilities to conduct ethnographic and other research to architect a user journey that anticipates and meets user needs.

For example, humans don’t speak in a straight line; we weave between ideas and points, switching between subjects. Or we make ambiguous references to previous topics, which require context of the ongoing conversation to make sense. Moreover, the order in which a person asks questions can impact performance, as more limited rules-based chatbots may frustrate customers by engaging them with dead-end topics. While humans can easily decipher these speech patterns, most chatbots currently require people to adjust their thinking and speech patterns to fit the bot’s abilities.

ACE+ looks to make virtual agents more dynamic, mining data from multiple locations and tapping machine learning to create an AI-backed tool that is continually learning and evolving. If the engagement is architected correctly, it can identify needs early and engage the best service for the application, ultimately facilitating more interactions across digital self-service channels.

"By decoupling the conversation flows and the conversation logic versus the natural language understanding — the AI that's interpreting the words and what your intent is — that has allowed us to recognize and carry through the context, thread, and the conversation," says Zinner.

As a cloud-based solution, ACE+ can be implemented quickly. For example, Accenture worked with government agencies across the country to implement ACE+ to meet the surge of inquiries — one thousand percent or more increase in inquiries — from the COVID-19 pandemic. In as little as a few days, the solution was operational, programmed to handle dozens of discrete intents. This allowed call center staff to focus on the more complex queries.

Ireland’s Office of the Revenue Commissioners was a pioneer in implementing the technology, showing the potential for using conversational virtual agents in the public sector. Working with Accenture, they developed a system encompassing over 200 unique dialogue steps, addressing 18 possible use cases and the capability to recognize 21 intents. In an initial pilot to provide 24/7 service, 55% of calls were handled start to finish by the virtual agent, and 70% of first-time applicants chose to engage with it when submitting their application.

But these tools are continually evolving, and as they improve, agencies will naturally want to continue improving their offerings. With ACE+, they can. A platform-based approach allows agencies to ensure they aren’t tied to a single technology and can move on to new tools as technology evolves and needs change. Importantly, organizations can centralize the knowledge within the system and plug and play new natural language understanding (NLU) services as they mature, eliminating the need to completely overhaul their offering every time an advancement is made.

"We're accounting for a rapidly moving marketplace by creating a vendor-agnostic platform that decouples specific components. As better technologies emerge, we can quickly integrate them into our existing knowledge base," says Zinner.

But even the most intelligent virtual agents won't be able to handle all requests — humans are still a key and essential aspect of the customer experience. For more complicated issues, agency staff can benefit from AI as well. Cognitive tools can augment customer service representatives’ (CSR) knowledge, suggesting actions and recommendations for human agents when interacting with consumers on calls and in live chats. This allows the CSR to focus more fully on the customer, using uniquely human skills like complex reasoning, creativity, and empathy to understand their issues. In contrast, the virtual agent uses its computing power to quickly find the most relevant information to respond with.

As agencies embrace authenticated digital experiences and deploy virtual agents across all of their customer service channels – voice, web, mobile and social – they are faced with the cost of supporting multiple solutions and ensuring they operate consistently. Investing in a single, enterprise-wide platform operating at scale helps them deliver a more unified and powerful customer experience. It also builds forward compatibility as new solutions can be readily integrated into the shared knowledge base as virtual agent technology continues to evolve rapidly.

“We’re looking to create one virtual agent brain to think holistically and create one virtual agent brain that knows how to interact through multiple channels instead of separate tools that drive separate experiences,” says Zinner. “AI is getting better all the time, and virtual agent solutions offer an exceptional opportunity to improve customer satisfaction at a sustainable cost.”