Part 1 of a two-part series.
Chatbots are taking the business world by storm. The popularity of this technology — which enables humans to converse with computers through their native language — parallels advances in messaging apps, the app ecosystem, artificial intelligence (AI), cognitive technologies and automation.
The hype around this phenomenon, however, may not be sustainable over time without a stronger business case and better near-term results. As the novelty wears off, users’ utmost concern will be how well the bot can get things done.
As the ecosystem takes shape, we see specialized chatbots emerging to address various user needs, including the following:
Customer service. Right now, most chatbots provide shopping assistance and product recommendations. Almost all major brands are said to want a chatbot of their own.
Scheduling. Buying plane tickets, booking movie tickets and setting up meetings — these are tasks that chatbots can easily complete. For instance, Alaska Airlines’ virtual assistant “Ask Jenn” helps with ticketing and customer service.
Status checks. These include apps that report on the weather, local events and the news — basically everything you need to know from the Internet. Examples include Telegram’s weather bot and CNN’s chatbot for news.
Entertainment. Bots can also deliver entertainment by offering random quotes, funny videos and jokes, such as Google’s Allo.
What Really Matters with Chatbots
While many chatbot designers focus on features and functionality that enable chatbots to mimic human responses and engage in natural, intelligent conversations, we believe the continued success of chatbots will not wholly depend on their ability to conduct a natural conversation. After all, the purpose of chatbots is not to build a relationship with the technology.
Furthermore, attempts to humanize chatbots can also give rise to what has been termed “the uncanny valley”: the theory that humans’ emotional response to a robot quickly turns negative when the robot appears “almost” human. The same theory can be applied to chatbots, many of which now swing between being an intelligent machine and acting human. The attempt to manufacture emotions — be it remorse, happiness, pride or frustration — can result in a negative experience for users who attempt to work with machines that try to emulate humans.
Another consideration with chatbot-based customer service is the effort it can take to explain complex problems. Chatbots can be very useful for problems that are clear, easy to articulate and straightforward to solve. Problems that are difficult to explain to a human at the other end of a phone are even more difficult to explain to a machine over chat. Being repeatedly asked to rephrase can result in user frustration.
We believe the role of chatbots is akin to an ATM, a technology that has never pretended to be intelligent or attempted to engage users with small talk. These machines did only what was needed: help customers make basic financial transactions at all hours of the day.
Over time, we believe chatbots will mature along the following lines:
They will become increasingly utilitarian. “Utility” in this context means something that’s considered so useful that people turn to it on a daily basis without hesitation. The focus should be on improving the competence of the chatbot. After all, a chatbot that fails when given a complex task is not sustainable.
First-contact resolution will become a key performance metric. Chatbots that provide solutions on the first instance, without the need to paraphrase or explain the problem in greater detail, will ultimately win. The ability to respond to user commands or queries in the best and shortest way possible will enhance the utility of the chatbot.
They will work to understand an individual over time. The ability to personalize responses will determine the quality of the customer experience. The chatbot must build a relationship with users based on their needs rather than delivering a generalized set of responses.
They will leverage data from a variety of sources. The bot’s ability to link data from both inside and outside the enterprise and create graph databases will be a key capability. The faster the chatbot accesses, processes and responds to data, the better the overall experience will be.
They will become more specialized. Users need to know what the chatbot does — and does well. Even top voice assistants such as Siri and Cortana do not excel at everything.
A “master bot” will be needed for bot management. As chatbots become more specialized and popular, managing them could become as overwhelming as managing apps. We believe a master bot will emerge that will act as a personal assistant, getting things done on behalf of the user, and even interacting with other bots to complete tasks.
The best bot experience is one in which humans get what they request and need. Bots will move slowly in that direction.
In part two, we will detail the key design elements for developing successful chatbots.
To learn more, please read “The Chatbot Imperative: Intelligence, Personalization and Utilitarian Design” or visit the Digital Business and Global Technology Emerging Business Accelerator sections of our website.