As I documented in my last post about the Opus Research census of bot deployments around the world, more than 1,600 companies have already tried their hand at implementing intelligent virtual asssitants. Chances are very high that your competition is doing so as well. Your firm is going to have to make some moves. As it does, it is good to follow some simple guidelines:
#1: Recognize the Redefinition of Self-Service.
The "botsplosion" represents nothing less than the redefinition of self-service on a global scale. Dating back to the 1980s, automated handling of service calls were carried out in the name of cost-savings. Success was measured in terms of call-deflection, capture rates, and the attendant reduction in the need for expensive, live agents to be involved in routine support functions.
Today, self service can be understood as, literally, serving oneself. Customers and prospects are taking control of their so-called journeys. A significant number of them actually prefer dealing with machines rather than other people. Contact center managers have long known that billing issues and promise to pay is best handled by automated outbound services. The same thing can be said for shopping and product selection, speedy resolution of service issues, restaurant reservations, service appointments, and a number of other activities that are both frequent and well-understood.
#2: Provide Your Customers with the Tools They Need to Take Control.
In many cases, companies have already baked the tools for intelligent assistance in their websites and mobile apps. Well-designed mobile apps provide instant access to key information, like account balances, order status, and personalized suggestions, recommendations or how-to articles that are highly relevant or context sensitive, in today's parlance.
The best mobile apps enable individuals to start a chat, make a phone call, or even make video contact with an agent, assistant, or advisor. Such in-app communications can start with a virtual representative, as often happens when a mobile customer moves to web chat, but it can quickly transfer to a live agent as required by the customer. This follows a phenomenon that you'll hear about over and over again: "human-assisted virtual agents."
#3: Implement Artificial Intelligence Judiciously.
Just as the term bot is generalized and invoked almost to the point of uselessness, the term artificial intelligence (AI) is overly broad and unhelpful, in the extreme. At a minimum, AI is used to describe computers that win at chess, Jeopardy! and Go and, therefore, are out to outperform humans in ways that we can't even predict yet.
In the world of Smart Customer Service, we're only interested in the aspects of general AI that can help agents or customers get questions answered, remediate problems, or otherwise improve their experience. Opus Research calls this intelligent assistance (IA) which is often referred to as intelligence augmentation in academic circles.
At AI's core is natural language understanding (NLU), which enables customers to use their own words to text or talk to an automated system. The proliferation of Apple's Siri, Amazon's Alexa, Microsoft's Cortana, and Google's Assistant have conditioned people to expect more from virtual assistants and, for the most part, they don't disappoint.
The other components of artificial intelligence include machine learning, predictive analytics, and knowledge management. When taken collectively, they enable intelligent assistants to constantly learn over time and improve their responses, anticipate each individual's intent based on their words, situations, and past activities, and, finally, to pull together information from a number of sources to add domain knowledge and provide consistent, accurate answers or recommendations.
#4: Provide a Single Truth Across Multiple Channels.
When done right, providing IA is a silo buster. Companies cannot afford to develop and maintain separate technology stacks for each customer touchpoint. Your e-commerce website, mobile app, Alexa skill, Facebook Messenger bot, IVR, and live representative should all provide the same answers to customers and prospects. The same will be true when the next shiny object capable of offering conversational access to self-service resources enters the scene.
This should be simple because there is, in fact, a fixed set of sources for the truth about each customer or prospect who contacts your firm. After they have identified themselves, you know where to look to find out whether their order shipped, you have a good idea whether their bill has come due, and you might be ready to recommend a rate plan that is an improvement over their current offer. Those responses should be identical regardless of whether they have called on the phone, talked through their TV remote, sent a Tweet at 3 a.m., or engaged in a conversation with your virtual agent.
#5: You Already Got This.
I'll close on a reassuring note: IA is well within reach. It is a matter of recognizing where true self-service has already been successful. Giving voice control to cable subscribers is a form of genius because they'll start with the simple stuff (changing channels or finding a movie) and then they'll get progressively more adventurous, looking to pay bills or upgrade their service packages.
Mobile apps are, likewise, the source of what I used to call "speechable moments." Delighting customers, in this context, means giving them videogame-like control over their sessions with your customer-care resources. Let them search for content and navigate their bills, and go in for the kill by getting a payment adjusted. It's already baked into your customer care infrastructure.
The challenge is to treat each new shiny object—be it a bot, a skill, an action, or a holographic representation of a robotic avatar—as yet another channel and to deliver consistently good, accurate, low-latency responses to your customers' requests on it.
Dan Miller is founder and lead analyst at Opus Research.