Chatbots – moving beyond the hype
In your organization’s shift to digital, the next milestone is customer interaction. To date, human-to-human communication has been the norm, but chatbots are becoming an increasingly viable alternative. Read why it pays to keep up with this trend, and what to watch out for.
What are chatbots?
Chatbots are computer programs that can hold conversations of varying complexity with real people. The term chatbot was coined in the nineties, when such programs were still very ‘robotesque’. It’s a bit understated for today’s more advanced Virtual Assistants. Using artificial intelligence and natural language processing, they can determine the context and the user’s intent, leading to a more life-like conversation. The most intelligent chatbots, called Virtual Agents, can handle really complex dialogs, processes and security protocols.
Why does your organization need them?
Traditional service desks are under pressure due to high staff turnover and related training costs. Meanwhile, your customers have less and less patience to wait to be served, and increasingly prefer self-service solutions. As chatbot technology advances, your competitors are already taking the plunge.
Why does your organization want them?
Chatbots work around the clock, are never ill and never resign. And when programmed well and with access to the right information they will never misinform customers. They’re easily scalable, allowing your organization to respond flexibly to changes in demand. They support routine processes, freeing your staff up for tasks where humans can make a real difference. Moreover, all their interactions with customers are stored, giving you a treasure of data about customer satisfaction.
Why does your customer want them?
Chatbots offer customers a low-threshold, round-the-clock platform for communication with your organization. Your customers can skip filling in complex online forms. They spend less time in helpdesk queues. And because their data is stored, they don’t have to repeat their story every time they call in again.
What are the risks?
Organizations may introduce chatbots too hastily, to perform tasks they are not yet a match for. Such organizations won’t get the return on investment they bargained for. Reputational risk looms, too. After all, customer frustration at failed interaction with chatbots can travel a long way on social media. And frustration is never far away, considering that chatbots cannot yet divine moods or decode sarcasm. Their self-learning skills can also backfire if they are fed undesirable information by customers. In less than a day after its launch, Microsoft’s Tay morphed into an internet troll.
Data security is paramount
The most serious risks, however, regard privacy and data security. A chatbot that asks customers for personal data, for example to apply for a new bank card, needs to be kept very safe from hackers. And what happens if your chatbot is approached by an external chatbot? Giving your chatbot access to your back-end system is appealing, as it greatly enhances its ability to solve customers’ problems, but the more access you allow, the more careful you need to be.
Success starts with groundwork
So how do you introduce chatbots successfully and safely? The first step is to do your research. Which tasks in your organization are suitable for a chatbot (and which are not)? Be realistic, and remember you can always scale up later. Next, limit your up-front investment by identifying the minimum requirements your first chatbot must meet. The choice of vendors is huge, ranging from giants like Microsoft and Google offering all-in-one solutions to smaller companies offering plugins. Look at the data security consequences in your organization and take any measures needed.
Training your chatbot
Before you can begin, you need to train your chatbot. That’s not so different from training employees: it requires time and investment, especially in the startup phase. The chatbot must learn what your organization does, how to answer questions and process requests, and how to access your knowledge and data systems. And when processes or products change, you need to retrain them, as you do with your employees.
Start small, and explain what you’re doing
When you’re ready to go, start small. Select an internal process for your first chatbot application, like allowing employees to arrange their own leave. For customers, start with a chatbot that can answer the Frequently Asked Questions on your website, for example. And don’t forget the communication side: explain to customers what you’re doing and why, and what’s in it for them. Manage their expectations regarding their first conversations with the chatbot. Collect data and learn and adjust as you go, the agile way. Step by step, you can move into more complex chatbot applications over time.
Future is around the corner
Although they have some way to go in terms of intelligence and natural language, chatbots are already a worthwhile tool for handling routine processes. But developments in this field are accelerating. Indeed, marketing guru Heather Pemberton Levy believes that “by 2020, the average person will have more conversations a day with bots than they do with their spouse.”