Computers are smart, but they can’t “chat” or “shoot the breeze” like your average human. Conversely, most people can perform any of the above in their sleep, but tracking down information in an online knowledge base or file store can border on the perplexing.
Bridging that communications gap starts with conversational AI and its component parts (e.g., natural-language processing, natural language generation, and dialogue management). Collectively, they can create a natural dialogue that helps people get answers to their questions. But the foundation for any effective conversational AI experience starts with having the right strategy, which begins with knowing the right questions to ask and steps to take.
Following are nine principles to a conversational AI strategy that fosters sustainable, useful, and efficient customer engagement.
- Know your customers. When developing a conversational AI strategy, start by asking the following questions: Who are we targeting? What do they need? How can we help them? What do we want them to know? Identifying these particulars will help crystallize your thinking in terms of the scope of the project, your audience’s need and how a conversational AI experience can meet them.
- Support numerous languages. If your target market includes users who speak a language other than English, expanding your Conversational AI (CAI) solution into multiple languages should be part of your strategy. CAI can easily, and quite accurately, detect a person’s preferred language, based upon their IP address, browser settings or HTML attributes. And while many people speak more than one language, the ability to interact with a chatbot in a user’s mother tongue goes a long way toward developing goodwill and fostering brand loyalty. Some CAI solutions use translation engines to respond to users in their preferred language, while others may require that responses be translated and fed to them in advance. If you need third-party support, for accuracy’s sake, look for a partner that can source data from native speakers.
- Thoroughly analyze your FAQs. Conversational AI can be a helpful platform for customers to find the answers they’re looking for. As such, be sure to conduct a thorough study of company FAQs. Doing so will reveal the most common issues and questions to which your conversational AI solution needs to respond. Furthermore, FAQs will highlight the named entities that a solution’s natural language processing algorithm should be trained to recognize. These entities are clues about the topic at hand, and give your conversational AI solution a cue about how to respond.
- Automate your workflows. By its very nature, an FAQ doc highlights the most common questions. As such, creating automated workflows for each of these questions and potential outcomes can help streamline employee and customer support, saving your company time and money. Mapping out the different outcomes for each question or issue may also help identify gaps where additional knowledge base content needs to be created, or where company policies need to be clarified in order to provide a consistent level of support.
- Create informative, understandable content. Your conversational AI strategy should also incorporate customer support content. For example, a truly useful CAI solution must be backed by informative, easy-to-understand knowledge base (KB) content. But don’t boil the ocean. Oftentimes, a person’s question or performance issue stems from one thing in particular. We recommend creating succinct chatbot flows that are backed up by equally focused KB articles, which include links to additional relevant content.
- Empower your service agents. Even the most well-developed AI application can’t answer every question. In fact, what often sets apart the most effective CAI experiences is knowing when, and how, to seamlessly handoff someone to a service agent, giving agents the necessary context of a service ticket, and empowering them to help resolve calls by surfacing relevant KB articles. At this point, the CAI application on the front end effectively becomes an augmented intelligence solution on the back end. Enabling this functionality requires a well thought out training process for entity recognition and metadata tagging of support content.
- Manage data rigorously. Tomorrow’s successes are built on the consistent business practices of today. With that in mind, it’s critical to develop a data management strategy and correlating disciplines that ensure you have access to quality datasets, which can help fine tune the performance of your AI application.
- Stay compliant. There are numerous potential points of exposure for user data, and there are compliance requirements from government/industry regulators on multiple fronts such as GDPR, HIPAA and SOC2. Furthermore, the use and storage of information related to areas such as identity, health, and online activity is being watched more closely than ever. As such, it’s essential that you develop a compliance strategy that’s robust, but which can also be easily adapted to unforeseen compliance issues.
- Track performance data and retrain your AI. An effective CAI application is not static. Rather, it is constantly being retrained to reflect a company’s latest programs, products and features, as well as changes in the vernacular of a language. Furthermore, it’s essential that companies track the resolution rate of their CAI, such as its ability to resolve a service ticket. Tracking its performance will help identify where additional algorithm training and support are needed.
Conversational AI has amazing potential to reduce costs and successfully resolve customer service calls. Achieving this success starts with a well-thought-out conversational AI strategy and complementary business and data management processes, which foster useful and efficient engagement.