The bot debates are on, and it’s déjà vu all over again. I remember the earlier version of the bot debates with voicemail and the IVR. Before voicemail, if someone wasn’t available to answer their phone, a cheerful receptionist would, and likely jot down a message on one of those pink ”While You Were Out” pads. When voicemail systems gained popularity in the ‘80s, they largely replaced live-answer receptionists.
There was debate back then about machines vs. humans, and whether it was cold to answer the phone with an automated voicemail greeting. To some degree the debate continues today, but voicemail became predominant -- not just because it was cheaper, but because it effectively eliminated the middleman. Voicemail made it possible to leave a direct, private, complicated, contextual, and even emotional message. Instead of “call me back,” entire conversations could become asynchronous. (Remember, voicemail was before email and text messaging became popular.)
A similar debate occurred with interactive voice response (IVR) systems. IVRs saved time and money by providing answers to common questions such as those regarding business hours or driving directions. They became more powerful with CTI and could respond to questions such as asking for the current balance on an account. Universities implemented touch-tone registration systems for students. The IVR was an impersonal bot, but that had advantages on certain types of inquiries such as setting a password or PIN code; it seemed odd to tell these things to an agent.
Now these debates are resurfacing with revised chatbot technology. A chatbot is similar to an IVR in function, but new AI technologies make them less programmatic. Clearly, they offer a number of benefits to customers and businesses, including faster and cheaper self-service, improved privacy, and 24-hour availability. Many demographics prefer self-service options but adding chatbots in the name of customer service is a murky proposition.
Though there’s no shortage of recent chatbot announcements, I’d like to explore this with three specific, recent announcements from T-Mobile, Moxy, and Google.
T-Mobile: In August, T-Mobile announced Team of Experts as a key differentiator in its un-carrier journey. When its customers (all customers, not just the best customers) call or message for service, they will be directed to a skilled team of highly motivated people. T-Mobile pledged the end of robots and phone menus, to reduce call center run-around (internal transfers), and made these experts available 24 hours per day. Two months later, the results are impressive. Its Net Promoter Score increased to an all-time high, up 60%. Use of T-Mobile’s messaging service for customer care increased 34%. Internally, agent turnover dropped 48%.
Those are impressive gains, but there are likely situations even at T-Mobile where a chatbot makes sense. It’s unfortunate that it had to vilify the chatbot technology to make its point. T-Mobile improved its customer service scores in a sector known for the worst customer service. The innovation was live answer by competent agents. The real question now is what’s next, and what’s the next innovation for sectors that already have these practices in place?
Moxy is the code-name for a new U.S. airline from David Neeleman, the founder of JetBlue. Neeleman describes Moxy as a “technology company that happens to fly airplanes.” His plan is to create a new airline that competes on both air travel and customer service.
As a founder of four successful airlines, Neeleman can’t be dismissed. He intends to deliver flights that are twice as fast for half the cost. Three key points to his plan are new fuel efficient planes, routes to/from second tier airports, and an app that empowers customers and lowers costs. An app-powered airline may sound dubious, but the idea of running a global taxi company without dispatchers may have sounded equally absurd only a decade ago.
Moxy won’t have customer service, at least not in the traditional sense. The Moxy app is being designed so that customers will never have to talk to the airline. The app is modeled after companies such as Amazon and Uber, for which live customer interactions are the exception rather than the rule. It will allow customers to do most things on their own, including changing or cancelling flights.
Live agents will exist, but they are not intended to be part of the normal workflow. The real innovation here is more flexible and powerful processes. Most interactions with airlines are self-inflicted by the airline. Moxy, for example, will proactively reschedule customers as a result of cancellations. Customers can accept the new schedule within the app without ever speaking to anyone. Clearly, Neeleman intends to rethink the workflow, and that likely means lots of AI and, yes, chatbots.
Google: This month Google started shipping its Pixel 3 smartphone with a new call screening feature. When a call comes in, a “Screen Call” button is presented alongside the familiar answer and reject options. Screen Call causes a Google chatbot to answer and collect information about the purpose of the call. As the caller speaks their responses, they are presented to the receiver as a live transcript on the smartphone’s display. The receiver of the call can interact with the chatbot via on-screen buttons such as “ask for more information” and “is it urgent?” At any time the call can be accepted, rejected, or marked as spam. Google claims that screening calls this way is fast, private, and uses less battery.
It’s a very clever use case that makes the chatbot available to consumers. It’s a logical addition for all smartphones and UC mobile clients. Although there are situations where it will be used with friends and family (is it urgent?), it will more likely be used to screen unfamiliar numbers. In other words, the Google screening chatbot will primarily interact with robocallers — let them chat about cake.
This is a fighting-fire-with-fire approach to managing robocalls, which are becoming pervasive. The next salvo will be the personal assistant bot that makes outgoing calls. Google demonstrated its Duplex technology earlier this year at Google I/O 2018. It’s a chatbot that performs common tasks, such as setting an appointment, by making calls. Duplex will begin rolling out next month to Pixel 3 owners on a city-by-city basis.
The Chatbots are Coming
Like it or not, the chatbots are coming. They are a key component of just about every major contact center vendor/provider’s roadmap and strategy. T-Mobile and Moxy are both leveraging the use (or non-use) of chatbots as a form of competitive advantage in completely opposite ways. Chatbots will be used to reduce costs and improve customer service, and in some cases accomplish both at the same time.
Though companies like T-Mobile are swatting automation, these technologies have a role to play. The maligned IVR enables valuable self-service, despite its ridicule. The design goals of improving customer experiences and reducing costs don’t have to be in conflict.
The issues of appropriate use of automated technologies are not limited to enterprise communications. The medical industry, for example, struggles with if and how AI-powered solutions for diagnosis should address bedside manner. These are complex issues, in part, because what’s deemed appropriate use of technology varies by age and generation.
The effectiveness and acceptance of bots may boil down to implementation. And, there’s no shortage of opportunity to do it right. Last July, Google announced Contact Center AI, and nine launch partners committed to upcoming integrated solutions (see “Google Embraces the Contact Center”). This month, Twilio launched its Autopilot chatbot service, aimed directly at business customers and offer a higher degree of programmability.
Two key success factors for chatbots are the workflow and the training. The more well defined the workflow, the more effective a bot can be. Many people prefer bots (and automation) in well-defined workflows such as placing/changing an order or scheduling an appointment. The training process is the Achilles’ heel for most implementations and will likely be the biggest differentiator among offers and implementations.
With Google arming the masses with chatbot technology, there’s going to be a whole lot of non-human conversation occurring. Today, we integrate disparate systems largely via APIs. This approach isn’t scalable, however, as it isn’t reasonable to integrate every system with every other system. Instead, chatbots will allow us to use voice as a universal API of sorts — just like humans do. Bots will likely do better talking to bots than humans as bot vocabulary and accents are more predictable.
Google agreed to make clear that its Duplex technology is robotic so as to not trick humans into thinking that it’s real, but it seems more likely that Duplex will be interacting mostly with other bots. We are moving to a time where every person and every business will have a bot.
Dave Michels is a Contributing Editor and analyst at TalkingPointz.