Chatbots are gaining favor by government agencies to improve service delivery.
“Before COVID-19, a few leading governments were dabbling in chatbot technology, using AI to address common resident queries,” according to Government Technology, which only polled state, city, and county governments. “In 2021, it’s hard to imagine government doing the people’s business without them.”
Federal institutions are also developing chatbots to provide 24/7 response to frequently asked questions. This trend is driven by changing customer expectations, according to the U.S. General Services Administration (GSA).
GSA’s Centers of Excellence (CoE) white paper, notes: “Smartphone AI like Siri has conditioned consumers to achieve more in short conversations with their most personal devices. Consumers have been taught to ask simple questions in conversational tones, and have the answer returned to them quickly.”
Get started by adopting a chatbot platform. Equally important is establishing an expert content and logic team to create conversational dialogue and to build “intents,” the building blocks of successful chatbots. Intent is the industry term used for knowing what users mean by their questions and how to provide the correct response path or link to the information they need.
Once the team and platform are in place, it’s a matter of testing and retesting the chatbot’s performance. No chatbot starts out as a perfect customer service agent. During this beta phase, it’s imperative to be transparent with users that the chatbot is in learning mode. In fact, Highlight finds that federal website users are generally understanding of a chatbot’s initial limitations, and many users are willing to take part in surveys and feedback sessions to help improve the chatbot. Use this opportunity to your advantage.
During the testing stage, host your beta chatbot on the help page and clearly state that the chatbot is under development and that feedback is appreciated. By placing the chatbot test below or above the fold, and trying out different attention grabbers and calls to action, you can control the number of users who are the first to experiment with the chatbot and provide feedback.
As development of the chatbot continues, be prepared to also share insight on website content and page logic improvements. Take advantage of this opportunity by working closely with the website and content teams. The more that the chatbot and website content work together, the more successful the chatbot will be.
Content remains king, says Ruxi Giura, a digital insurance specialist for the National Flood Insurance Program at FEMA, writing for Digital.gov.
“Chatbots can help users find information faster, but they are not a replacement for answers to frequently asked questions on your organization website. Chatbots require good content that can be easily parsed, and that content must be created and maintained by humans.”
Once the chatbot has achieved a significant level of proficiency – getting users to the right information over 80 percent of the time, for example – the chatbot can then populate and follow visitors throughout the website. Federal agency websites often cover many disparate topics, so you may need to concentrate on chatbot logic for one subject or service area at a time. Different intent logic or even different chatbots may be required for different topics.
Chatbot development is a highly iterative process. Don’t, as the adage goes, allow perfection to become the enemy of the good when it comes to chatbot development.
Chatbots “are often structured for triage, the weeding out of people whose questions can be answered easily so that call-takers can focus on people whose questions will take more effort,” writes Ben Miller in Government Technology.
You will have to accept that there will be times when the chatbot doesn’t have the right answer. That’s when you apologize for the chatbot’s limitations and direct the user to the customer service team.
By establishing a solid foundation of intent logic, and constantly experimenting, you will continue to gain data that can be used to grow your understanding of user questions and needs. Once your baseline chatbot structure is in place, you then can work on continued improvements with the use of artificial intelligence and natural language processing.
Federal agencies will continue to advance chatbots to provide on-demand online support backed by a team of human agents who will handle more complex questions. Those who have already started will be the first to gain the advantages of more sophisticated chatbot tools.
Author: Barry Lawrence | Senior Communications Consultant