2019 approaches and the C-word, “Conversational” is everywhere. It has become the ubiquitous antecedent on websites, product collateral and analyst reports touting “Conversational AI”, “Conversational Marketing”, “Conversational Analytics” and (my personal favorite) “Conversational Commerce”. Its importance already extends beyond Contact Center Operations, Digital Transformation Initiatives. Conversational resources are the foundation for measurable improvements in Customer Experience (CX), Self-Service and, more broadly, Process Automation.

So Many Channels and So Much Data

Beyond 2019, enterprises must support chatbots, message bots, Alexa Skills, Google Actions and the inevitable advent of new conversational interfaces in cars, kiosks, retail stores and homes. In 2019 brands will invest tens of billions of dollars to make sure that conversational resources support better customer experience and enhanced employee productivity. As the saying goes (or should go) “with investment comes responsibility.” Here are some guiding principles defined by identifying the first-order issues that conversational advocates must tackle and, by association, the hard problems that the most thoughtful folks in the business are investing in:

• Recognize that Conversational AI is *not* a thing: AI is a highly imprecise term that needs to be narrowed down to apply to purpose driven applications and use cases. In doing so, practitioners find that some elements are more important than others. Natural Language Processing (NLP) is core to recognizing the category of a contact and start arriving at the intent of an individual prospect or customer. Data scientists that I have spoken with consider NLP to be a subset of Machine Learning (ML) which is a computer discipline that has become vital for a company to constantly improve responses and recommendations based on natural language input. With the number of #VoiceFirst devices growing exponentially, Automated Speech Processing (ASP: speech recognition, text to speech and voice biometrics) are growing in importance as well. Finally, investment in various flavors of “analytics” (primarily speech, text and predictive) are proving to be quite vital for design and deployment of an Intelligent Assistant that can consistently predict and rapidly respond to a customer query or direct instructions.
• Listen. Don’t surveil: The implementation of General Data Privacy Regulations (GDPR) in Europe is having a global ripple effect, and rightfully so. Brands have become over-reliant on data about their customers or prospects that was aggregated, acquired or otherwise captured, ideally, to support rapid and accurate response to queries or instructions at one of many touchpoint but, more often, to support targeted outbound emails or advertising placement. The mission for NLP, ML, ASP and Analytics in 2019 is to transform companies into better listeners. This does not mean leaving an open mike on in board room, bedroom, car or kitchen, it means investing in the resources that make sense of what people say after they use their wake up words, initiate chats or call a contact center.
• Take advantage of every API out there: In 2019, more than ever, it will pay to have connections. The giants of cloud computing and e-commerce have invested billions in their cognitive resources. APIs for Amazon LEX, Microsoft LUIS, Google Dialogflow and IBM Watson are proving quite popular and members of a cottage industry, exemplified by start-up zapper.com, make it easier to shop for “connectors” that integrate popular apps and processes, like email, scheduling, ecommerce, knowledge management, CRM and payment management, into Intelligent Assistance workflows.
• Make room for humans: Dissatisfaction with IVRs in customer care contact centers gave rise to the “Get Human” movement, as individuals tried reflexively to opt out of automated handling. By contrast, in the Intelligent Assistance domain, early implementers (circa 2013) were pleased to observe that their chatbots and IAs were able to deliver high automation rates at higher levels of customer satisfaction, measured in Net Promoter Scores (NPS) and CSAT surveys. This raised a second set of concerns as humans in the form of live agents felt that they might be training their robot replacements. In the ensuing years, practitioners have defined deployment strategies for Intelligent Assistance that support the goals of customer care employees and create new roles for employees in the creation and refinement of IA offerings. They work with solutions and platform providers that support “zero code” service creation and they enlist departmental employees (rather than techies) into efforts to refine responses and reinforce Machine Learning.
• Use IAs to KISS, not!: “KISS” is the classic acronym for “Keep It Simple Stupid”. But that expression is both stupid and insulting. Customers aren’t stupid and their queries are seldom simple. Brands can choose one of two distinct strategies for using IA technologies to support their customer care or conversational marketing objectives. First is the Answerbot. They are direct descendants of the “one-and-done” chatbots designed to understand natural language input and provide an answer based on a relatively small set of possibilities. But nothing is that simple. in 2019, IAs have the capability to understand a broad set of utterances or text input from customers. Conversations can include multiple “turns”. or back-and-forth interactions with customers or prospects. Answers or recommendations are informed by relevant sets of “Big Data” (both structured and unstructured) that span product catalogues, customer records, chat transcripts and, realistically, the sum total of human knowledge that has been made available to search engines whose spiders crawl the Internet.

Privacy, Simplicity and Precision will be the watchwords of 2019. Firms will succeed by incorporating elements of Artificial Intelligence into their customer care and digital marketing infrastructures to do the sort of power listening that the combination of technologies make possible. Smart speakers, smartphones and personal virtual assistants are conditioning people to ask for exactly what they want in their own words.

(picture credit: Akuland Group)

Categories: Intelligent Assistants, Articles