Why customer service matters for government — and how AI will help

Government agencies have their work cut out for them; consumer expectations have never been higher. Digital interactions must mirror or exceed in-person ones, and consumers expect high levels of customer service and efficiency. Consumers are accustomed to, say, quickly returning the wrong size phone case they ordered online, or cashing a check via their mobile app. From a consumer’s perspective, why wouldn’t the quality of customer service and digital interactions be just as high — or even higher — for issues that have bigger impacts on their livelihoods and well being, such as applying for disaster assistance, signing up for food benefits, or connecting with the right healthcare services?

In this digital, customer-first reality, 90% of consumers expect an immediate response to a customer service issue. This is a high bar, and perhaps more challenging for the public sector to reach. For example, a McKinsey study found that the public sector is ranked lowest of 10 industries for customer satisfaction, and only 8% of Americans say the US government is responsive to the needs of ordinary Americans, according to Pew Research.

Behind the scenes, providing consumer-level customer experiences for public sector can come with outsized obstacles.

Unlike in other industries, public sector must serve all people, not a market segment or niche (which often have off-the-shelf technology tools specialized for them). Because of this, agencies must account for customers with accessibility challenges, lack of digital access, geographic distance, language barriers, and generational divides.

For citizens, the transition to digitized government brings the new challenges of overwhelming data, options, and digital portals to access help and services. As technology becomes the default for government interaction, call centers are usually the starting point for people who lack the skills or access to navigate a sea of mobile applications and government webpages. 

Even the role of a call center is becoming more complex. Just finding the right entry point for a new caller is difficult. Being a call center representative is like being the receptionist for dozens, if not hundreds, of different businesses, all with their own policies and procedures. We've all had the experience of being redirected five times by a call center, impatiently waiting to reach the right person with the right information. It isn't fun.

From a customer point of view, most time spent with support is verifying your identity and then telling each successive representative what your issue and history is. Getting personalized support is frustrating, because at each turn it's like the government has forgotten everything about you, when realistically they should be the system of record for the problem that needs solving.

Today’s reality is that back-end processes, IT architectures, and data sources are not unified, meaning that customer service agents — and citizens looking for self-service — cannot draw from a complete picture of information. Compounding the issue is the fact that government agencies often work with external mission partners, spanning public-private partnerships, charitable organizations, and other government agencies at the national, state, and local level. All of these partners have their own sources of data. 

Customer support, armed with AI comprehension of interaction history and customer details, has the potential to streamline the complexity of government. Imagine if agencies could use all available data to resolve issues and achieve the customer experience revolution called for in our priorities and frameworks. You shouldn't have to tell the government “we already tried that and it didn't work.” The AI should know that and suggest a viable solution to the customer support agent. 

The good news is that generative AI is poised to help government agencies deliver faster, more personalized customer service — without putting additional pressure on your internal teams.

Consulting firm BCG estimates that the US public sector will collectively experience $519 billion in productivity gains from generative AI by 2033. In the realm of customer support, generative AI can boost both response time and relevance. The key is implementing these tools responsibly, securely, and with the right contextual data — and this is best done through retrieval augmented generation (RAG).

RAG ensures that generative AI outputs are contextual and relevant to your specific mission and agency. This specificity is enabled by adding a context layer between your agency’s proprietary data and large language models (LLMs), serving as a vital connection between data and generative AI responses. This context layer consists of highly relevant, real-time search results pulled from your own data. 

After these search results are filtered into a generative AI application, you end up with better, human-like generative AI responses (in question-answering or chatbot experiences, for example) at a lower cost. Without this critical context (and a way to search and access it), an LLM will only provide minimal value, as it pulls only from the data it’s been trained on (often outdated, publicly available data sets).

>> Learn more about RAG

The Elastic Search AI Platform harnesses the power of generative AI and search technology to optimize self-service customer support and agent experiences. Use your agency’s proprietary data, knowledge base, and process documentation to provide precise question-answering and recommendations, while maintaining document-level security and keeping costs low. 

Here are three benefits of using Elastic’s Search AI Platform at your agency’s call center or on your website: 

1. Increase access to programs and information via self-service. The US government estimates that $140 billion in potential benefits go unused each year because of “outdated or complicated processes.” Generative AI has the ability to streamline these processes and more directly connect citizens with information and services that they’re looking for. And because 81% of customers attempt self-service before reaching out to a representative, setting up self-service avenues, especially digitally, can take significant pressure off your customer service representatives and create faster ways for people to find what they need directly, especially outside business hours.

2. Reduce support time. When citizens choose to interact directly with an agency via a help desk or call center, generative AI can significantly speed time to resolution. With relevant, contextual search and retrieval, customer support agents can resolve support queries faster by surfacing answers and information that are personalized for individuals and their specific needs, instead of spending time starting with generic documents and resources.
   
   AI for customer support can help call center agents immediately find an answer, without needing to pass a customer off to a handful of other people. With real-time access to call scripts and best practices, AI can act as a subject matter assistant for the first person a customer talks to, making them instantly knowledgeable. AI can also help support employees switch context faster and expand to wider topics more easily than a busy call center worker could on their own. 
3. Better understand customers through unified data. Store and unify customer and search data on one platform in order to deliver analysis and insights that your agency can use to improve services and access to them. 
   
   When customer support agents are armed with an AI search view of a customer's history of interaction, they can get instant AI access — including summarization and recommendations on how to resolve not just based on the current call but the entire support ticket and history. You shouldn't have to tell the government “we already tried that and it didn't work.” The AI should know that and suggest an alternative for the customer support person to suggest.
   
   From a macro level, agencies can use unified data trends to optimize services and information. You find out what the most searched-for questions are on your agency’s website or what geographic locations these queries are coming from. Ultimately, these patterns can help steer future strategy and direction, aiding your agency in delivering better targeted programs and support or directing content to specific demographics.
   
   As noted on performance.gov, “Improving the public’s trust in government happens transaction by transaction; by understanding our customers’ needs through gathering feedback, conducting research, sharing insights, and testing new approaches to ensure we are designing with our customers every step of the way.”

Elastic’s Search AI Platform is easy for developers to implement, with full search capabilities for building AI apps and RAG workflows, document-level security, a production-ready vector database, our ELSER pre-trained retrieval model for more relevant semantic search results, and E5 (multi-language) support. Elastic’s open approach allows your team to integrate your data, securely, with your own or third-party transformer models. Contact our team today to walk through your approach and discuss specific next steps.