How conversational AI can deliver exceptional citizen services

Customers called government agencies about taxes, travel, student aid, Medicaid benefits and other services more than 194 million times in fiscal 2018 and visited government websites nearly 1.7 billion times, according to a recent Partnership for Public Service report. While there are pockets of excellence, the federal government, as a whole, too often lags the private sector in customer satisfaction as measured by major customer experience indices. And with continual improvements from commercial providers setting even higher expectations for what constitutes good service, customer expectations for everyone, including federal agencies have been raised. 

Our experience has found that customers expect to interact and transact around the clock via frictionless, connected experiences spanning every possible channel. They also want providers to immediately know who they are while also respecting their privacy. Furthermore, they expect providers both to anticipate and be responsive to their unique needs. And they want organizations to engage them in more relevant and personalized ways.

While there’s no question that the federal government faces unique constraints in serving the public, it still has opportunities to change the service game by adapting commercial innovation and emerging technologies to its unique requirements. 

Some of the most promising include artificial intelligence solutions for customer service, which are growing in adoption and maturity. Powered by machine learning with underlying natural language processing and intelligent automation capabilities, these conversational AI solutions use chatbots and digital agents to support every stage of the customer experience -- from engaging citizens and answering questions to processing requests and optimizing decisions. 

Agencies that implement conversational AI technologies across one or two touchpoints can improve performance but are unlikely to truly transform customer service. A better approach: combining and integrating these capabilities through a conversational AI platform to deliver a more integrated, end-to-end experience. In doing so, federal leaders can realize efficiencies through automation while delivering a more convenient, intuitive and desirable user experience. This approach enables both innovation and scalability, as agencies can both quickly add and more easily support new technologies while maintaining future portability.

This approach is serving as the cornerstone of customer service within many high-touch commercial sectors, including banking, hospitality, telecommunications and insurance. It is helping these enterprises deliver a highly differentiated, individualized user experience, at scale, for every customer engagement. Some have experienced a 30% reduction in operating costs and vastly improved their customer satisfaction scores as a result.

Voicebots deliver returns in Ireland

Commercial entities aren’t the only organizations that are seeing benefits. Ireland’s Office of the Revenue Commissioners (Revenue) – which receives three million calls a year seeking assistance – has realized compelling results that show how an AI-enabled platform can make customer service less taxing for all stakeholders.

While most calls to Revenue are not especially complicated, they all require time, attention and friendly assistance. CIO John Barron recognized the potential to use AI, especially natural language processing, but also the need to implement these tools effectively. Instead of trying to address every potential scenario, he focused his first pilot on a subset of calls related to tax clearance. The goal was to prove the technology was effective, including customer receptiveness and satisfaction with the new approach.

Barron and his team took advantage of a conversational AI platform to reduce time-to-market to just a few months while providing future flexibility. As part of the development, experts in both AI and voice experience design collaborated to create a truly human and intuitive experience. 

Building on their analysis, a solution was quickly developed encompassing more than 200 unique dialogue steps, addressing 18 possible use cases with the capability to recognize 21 intents. It drew together speech detection, text-to-speech and natural language processing technology. Early on, the team realized to the importance of crafting very targeted questions for agents to ask, in order to ensure the customer was guided down the right path. 

The first bot went live in June 2018 and quickly exceeded expectations, handling more than 2,000 calls in the first six weeks. Its performance showed that it was clearly meeting user expectations, with 50% to 60% of calls handled from start to finish by the voicebot and 70% of first-time applications who engaged with the voicebot submitting an application via the service. Beyond that, three-quarters of callers with active tax clearance were able to retrieve an Access Number, and just 10% of calls were transferred to a human agent due to failure to understand. From the outset, this voice-first pilot answered Revenue’s needs while delivering improved customer service, reduced costs and increased efficiency.

Achieving transformational success with conversational AI takes time, but real progress can be achieved in the near term. We recommend agencies start with a strategy to help prioritize impact, feasibility and desirability -- and use a platform-based approach to pursue both quick wins and longer-term strategic priorities.