It would be poetic justice for 2020 to be the year that everything in Conversational Commerce comes into focus; but that is too much to ask. Still, several positive trends are afoot as we enter the second decade of the millennium and should result in very positive outcomes ten years hence.

Here are the Top Five:

1. Conversational AI adoption is accelerating and is “use case driven”.
2. Past investment in stand-alone chatbots can join IVR and mobile to support Conversational Commerce.
3. The #1 challenge is building a “Single Source of Truth” to inform agents, chatbots and voicebots with the consistently correct answers or recommendations at scale.
4. Intelligent Authentication is more important than ever.
5. Brands will benefit from optimizing both customer and employee experience by thinking of them as workflows.

Below are the details.

Moving from Early Adopter to the Early Majority

First and foremost, market requirements have fundamentally changed from those of “early adopters” to today’s “early majority.” The former were willing to pay a premium for professional services while hiring their own staff of computational linguists, Natural Language specialists and conversational user interface designers. The latter are more frugal, pragmatic and mindful of business objectives.

Today’s implementers take a “do-it-yourself” (DIY) approach, but do so without a willingness to hire specialized staff or contract for exorbitant subcontractors. Their subject matter experts (SMEs) come from the ranks of long-standing customer care agents or supervisors. Their design chops are honed on tools that resemble Microsoft’s Visual Studio or Visio. Their patience is as short as their budgets are limited.

Speaking of budgets, funding sources have shifted as well. At large enterprises, specialists in customer experience (CX), digital transformation and omnichannel strategies have wrested both purse strings and vendor selection from silo-supporting decision makers in IT, contact centers or mobility. Mid-market companies never had the luxury of a multi-disciplinary approach. Funds come general operating budgets and are earmarked for marketing or sales optimization. Solutions are designed to accomplish a specific business objective, therefore it is use-case driven. Today, they are most likely to be defined by their long-standing providers of communications platforms as a mandate to infuse contact center, communications or (more recently) CRM platforms with the proper amount of predictive analytics or artificial intelligence to support chatbots, augmented IVRs or other flavors of contact management or sales acceleration.

Shifting from “Chatbots” to Conversational AI

Three years ago, Opus Research noted that investment in what we call “Enterprise Intelligent Assistants” (AKA “chatbots”) were queries coming from top management. This gave rise to development and deployment of hundreds of thousands of “chatbots”. Too many of them specialize on a narrow set of functions like form filling or Q&A at a departmental level and are not integrated with counterparts in other divisions or functional areas. Queries from top management have rightly evolved from “What is our bot strategy?” to the slightly more sophisticated “What’s our AI strategy?”

The distinction reflects subtle evolution from the belief that automated virtual agents, AKA “bots”, could be quickly trained to handle the bulk of repetitive interactions with prospects and customers. These digital employees brought the dual benefit of lowering operating expenses through automation while boosting morale among customer care professionals by engaging them only when their digital counterparts were unable to resolve an issue. In this idealized IA Utopia, the very nature of work for customer care agents is fundamentally changed from dull, repetitive activities to much more creative and challenging handling of more complex problems that are result in more meaningful and gratifying engagements with customers.

Experience has been a great teacher. The early adopters now understand that they must reel in spending on a multiplicity of proof of concepts and special purpose bots and take a enterprise-wide approach to improving customer experience in a conversational model, over time and across multiple channels. Adding more Intelligent Assistants is an expensive, managerial nightmare. What’s needed is company-wide, coordinated management of Conversational resources.

Providing Conversational Intelligence to Inform Everything and Everyone

This will prove to be the make-or-break factor for successful Conversational Commerce. The bulk of chatbots and virtual agents in service today are trained by using stored chat histories or transcribed phone conversations. Initiating a new IA starts with aggregating and ingesting voluminous amounts raw material from historical conversation and then dedicating both personnel and raw computer power to the processes of categorization, or tagging, those chats to identify the topics that have the highest impact on key performance indicators.

That activity is followed by the equally resource-intensive process of matching intents to desired outcomes. For rudimentary FAQs (frequently Asked Questions), the processes are simple. Subject matter experts, among employees or fellow customers, essentially vote to identify the best answer. The system can support constant learning by asking recipients whether they were provided with a satisfactory answer as part of an ongoing process of constant improvement.

In the static world of FAQs, conversations are “one and done” interactions that involve a single “source of truth.” Outside that world, providing Conversational Intelligence is one of the biggest challenges for enterprises of all sizes. Correct answers or recommended actions are the products of a myriad of processes taking place elsewhere in a company’s IT system or outside on the World Wide Web.

In retailing, for instance, a customer’s query may start with asking general advice, like “what makes a good gift if I’m going to be a house guest during the holiday?” That may be routed to answers from a customer forum, or it may be routed to a live agent (depending on a customer’s status in a so-called “Loyalty Program”). Let’s say the initial query is resolved with a recommendation for a floral arrangement. The next step would be to ask and answer a series of questions about price range, color options, seasonal promotions, and whether they might prefer a plant versus flowers. To inform this conversation, the automated virtual agent must know what options are available for the location under discussion, what is in stock in different retail outlets and other relevant, yet dynamic information. Then, when a decision is made, the inevitable progress to a checkout system must be accommodated in a conversational mode.

Authenticating End Users Supports CX, Security, Personalization and Trust

Authentication is advancing in parallel with advancements in AI and automated Intelligent Assistance. Before a brand picks up the phone, engages in a chat or begins to exchange messages, it needs to establish confidence that the individual at the other end of the line is who he or she claims to be. That is fundamental to establishing the level of trust that enables an intelligent assistant to tailor responses that reflect an individual’s preferences and personal history or, eventually, accept payment or instructions regarding personal information.

Opus Research coined the term Intelligent Authentication (IAuth) to capture the idea of continuous, frictionless authentication of individuals in order to support Conversational Commerce – over time, across multiple devices and involving multiple channels. Without a doubt, it has been the hardest concept for us to promote largely because there has always been tension between security professionals whose emphasis is on preventing access to imposters even it might mean rejecting genuine customers and their counterparts in marketing or CX who want to lower all barriers to commerce, even if it means letting an imposter in the front door (infrequently, but on occasion).

IAuth is the answer to both of these constituencies, especially if the prevailing engagement model is conversational. As we enter 2020, password management is out of control. Hacker attacks don’t stop with gaining access to personal data on hundreds of millions of people from credit card issuers or large retailers. They now like to show off by demonstrating that they used password vulnerabilities to gain access to home security systems, including, most dramatically, talking to an 8-year-old through the supposedly comforting Ring Nanny Cam in her bedroom.

Security experts quoted in popular media and Web sites now tell individuals “to pick unique, strong passwords for every service and device that you currently use.” That just isn’t possible or practical. Biometrics, like voice, face or fingerprint, should gain traction as intelligent endpoints, including nanny cams, smartphones and smart speakers, proliferate. Enterprises will be direct beneficiaries as people become accustomed to strong authentication taking place passively, “in the background” as they carry out their normal activities.

Evolving into Conversational Process Automation and Workflow Assistance

Adoption and deployment of both Intelligent Assistants and IAuth will be use case-driven. The number and span of those use cases are already expanding beyond customer support in contact centers to include sales acceleration, marketing support and advertising targeting. When Intelligent Assistants are assigned to join developer scrums or listen in on conference calls to provide summations and suggest next steps, the result is what Opus Research terms Conversational Process Automation (CPA).

CPA is amounts to machine-aided human productivity. In the context of customer care, it involves the recognition and acceleration of a customer’s “workflow” as he or she progresses from search, to discovery, to consultation, to product selection, to purchase. From the customer care agent’s point of view (for both automated and live agents) workflows span authentication, intent recognition, issue resolution, upselling. cross-selling and “compliance” with prevailing regulations all the while.

Customer workflows are an underappreciated side of CX. Companies that have pursue strategies for “Journey Orchestration” are aware of the need to manipulate customer activity based on status (state), preferences and past activity. That is the beginning of recognizing, simplifying and reinforcing existing customer workflows. Workflows can also be a unifying concept when it comes to matching agent activities (both live and virtual) with those of the customer. Plus fostering CPA and “workflows” is in total concert with investment and activity surrounding Robotic Process Automation (RPA), which is a term that has round new life as enterprises endeavor to leverage past investment in IT infrastructure and cloud-based services.

Happy New Year and New Decade.

Categories: Intelligent Assistants, Intelligent Authentication, Articles