Want Your Own Chatbot? Don't Move Forward Without Taking These 3 Steps
For years, tech enthusiasts, innovators and entrepreneurs have painted a bright future for chatbot technology. But the road to mass adoption hasn't been without its speed bumps.
Facebook's massive Messenger bot rollout a few years ago left a lot to be desired, and we've seen myriad me-too companies flood the market with less-than-stellar technology. Still, when they're done well, chatbots can be amazingly innovative. For instance, Amazon's Alexa has the ability to give users personalized gift recommendations. And recently, Google unveiled a chatbot that was so lifelike it actually fooled testers into thinking they were speaking with a real human.
Thankfully, entrepreneurs don't have to get on Google or Amazon's level to drive business results with chatbots. But implementing them the wrong way will do more harm than good. That's why it's absolutely crucial to understand the pros and cons of chatbots before unleashing one on your customers.
The ups and downs of chatbots
Chatbots are fast, accurate and increasingly able to anticipate users’ needs. And thanks to machine learning, their messages can become even more personalized over time. At Hatchbuck, we use chatbots on our website and find they’re a quick way to capture more traffic with a lower barrier to entry. Using a chatbot, we’re able to both answer customer questions and capture more of the qualified traffic making it to the site.
Of course, chatbots aren’t useful for every business model. Humans are still necessary for more complicated service-related tasks. Some companies, for example, provide in-depth consultations that just can’t be accomplished with a simple template.
Chatbots can also struggle to hold people’s attention. According to BotAnalytics, approximately 40 percent of users don't get past the first text, while another quarter usually abandon ship after the second. When they do hold customers' attention, they sometimes get lost in a conversation due to their inability to understand human nuances, such as sarcasm.
Preparing for launch
For us, deciding to launch a chatbot wasn’t a frivolous decision. It was a process that took several months to execute. Here are the three steps we took to get there:
1. Define a purpose and choose a chatbot.
Chatbots come in two different flavors: rule-based and advanced. The rule-based variety are quicker to develop but also rudimentary in their capabilities because they can only respond to specific commands (think “Alexa, play Taylor Swift"). A pizza place that just needs to know customers' addresses and toppings choices can easily get away with using a rule-based bot.
Advanced bots, on the other hand, leverage machine learning to understand natural language and participate in more complicated conversations. As they collect more data, they become more sophisticated. EMarketer reports AI can manage about 65 percent of inquiries at first, but after awhile, the robot can manage up to 80 percent.
Regardless of what type of chatbot suits your business model, ensure that it actually serves a purpose. Take KLM Royal Dutch Airlines' chatbot as an example. BlueBot, as it's called, helps customers book tickets, alerts them to any gate changes and updates them on the status of their flights -- it handles more than 16,000 interactions a week.
2. Don't dupe customers into thinking your chatbot is a person.
As a startup, your chatbot likely won't be as advanced as Google's, but it should at least be able to understand the flow of the conversation. Accomplishing this natural language understanding isn’t an easy feat, but it shouldn’t be unleashed on your customers without it.
Furthermore, even if your bot somehow sounds as natural at Google's, I wouldn't recommend trying to fool customers into thinking it's a human. When visitors enter our site, we're sure to let them know right away that a bot has greeted them. And if the conversation requires it, don't hesitate to let a human representative take over. In fact, a joint survey by Drift, SurveyMonkey Audience, Salesforce and myclever found 43 percent of people still prefer to communicate with humans over bots. So though a chatbot can be helpful, it should not be treated as a total replacement for human interaction.
3. Give your bot a performance review.
Juniper Research found that chatbots are expected to save businesses $8 billion by 2022. This is because person-to-person interaction costs seven times as much as an automated response, according to a survey by ContactBabel.
To reap these cost-savings benefits, the bot needs to perform in three key areas: It should comprehend what a user is saying and respond without error; it should deliver meaningful, personalized information that keeps users engaged; and it should communicate this information as quickly as possible to facilitate a natural conversation. As you're evaluating the performance of your bot, make sure to measure it based on these key performance indicators.
Chatbots aren’t a one-size-fits-all tool. They require a deep understanding of how customers are communicating with your brand. But while they take some work and research to implement, using a chatbot has made a profound impact on our business.