If you’ve ever started a sentence with “Alexa…” or “Siri…”, you’ll know that we humans are now well used to communicating with machines through natural human language. Chatbots are underpinned by the same technology as voice interface systems like Siri, but instead of responding to spoken commands, chatbots interact with users via a written chat interface, such as Facebook Messenger or a web-based application.

Like many other AI-driven technologies, chatbots have become a key technology trend. Today, businesses big and small are using chatbots to interact with their customers, drive sales, solve user problems, and more. In fact, chatbots are being used in a wide range of business functions – customer service, sales, marketing, tech support, HR – across a surprisingly diverse range of industries. But, impressive as the technology can be, there are some pitfalls and challenges to be aware of. In this article, I look at some of the most common mistakes businesses make when deploying chatbot technology.

Mistake #1: Not using chatbots in a strategic way

As with any technology trend, there must be a strategic reason for introducing the new technology to your business. Adopting technology for the sake of it can be costly and counter-productive. Rather, any new technology has to deliver real value for your business, whether that’s by improving the customer experience, answering queries more quickly, providing a more personalized experience on a large scale, or whatever.

Philippines-based company Globe Telecom shows how chatbots can deliver serious strategic value. The company deployed a Facebook Messenger chatbot to help ease the load on its call center. The result? Call volume decreased by 50 percent, while customer satisfaction increased by 22 percent. What’s more, employees were 3.5 times more productive.

Mistake #2: Using chatbots when you should have used a human

It’s true that chatbots are now capable of impersonating humans pretty well and holding impressively natural conversations. But there are times when only human interaction will do. Unless you want to alienate your audience, you’ll need to think carefully about which tasks are suited to chatbots and which are best left to humans.

Consider the use of chatbots in HR. Vera is just one of many HR-focused chatbots that have hit the market in recent years. Vera is a recruitment chatbot, capable of interviewing as many as 1,500 candidates a day, by telephone or video, and sending personalized follow-up emails. The tool is designed for companies that need to recruit a lot of employees, quickly. As such, it’s proven particularly useful for retail and distribution companies – IKEA Russia is one Vera customer.

Clearly, recruitment chatbots could become a valuable part of the recruitment process, helping human recruiters save time on first-round interviews and taking some of the leg-work out of communicating with candidates. But that doesn’t mean they should replace human recruiters altogether. The final decision on whether to hire someone should always be left to a human.

Mistake #3: Underestimating the potential for brand engagement

One error that many businesses make is to conflate chatbots with self-service offerings like online FAQ pages. Chatbots offer so much more than just helping people solve a particular problem or complete a certain task. Because they’re conducting a real conversation with the user, chatbots provide a genuine sense of interacting with a company or brand – in a much more meaningful way than, say, accessing information on a web page or completing an order online.

In other words, chatbots can seriously boost engagement. That was certainly the case for National Geographic, who created an Einstein-inspired chatbot for a new show called Genius. The bot replied to interactions like Einstein would, which definitely kept people coming back for more. Conversations with the bot lasted an average of 6–8 minutes and resulted in 50 percent user re-engagement.

Mistake #4: Not communicating as your audience would

Any successful chatbot must be designed with the target audience in mind and be able to chat in the same way as the real-life people it’s communicating with. The Whole Foods Facebook Messenger does this particularly well. The chatbot dishes out recipes, cooking inspiration, and product recommendations and drives users to the company’s website. So far, so normal. What’s special about the bot is it lets users search for recipes using emojis – thereby reflecting how people communicate in real life. Love them or loathe them, emojis are now firmly part of everyday communication, and they’re here to stay.

Mistake #5: Not improving the service over time

Chatbots are underpinned by AI. And the thing about AI is it’s constantly getting smarter. We need only look back at some of the high-profile, toe-curling examples of chatbot fails to see how far the technology has come. (Microsoft’s racist, Holocaust-denying Twitter bot Tay springs immediately to mind. Although, to be fair, it was Twitter users who taught Tay to be racist.) Therefore, you must be prepared to tweak your offering over time in line with user feedback and technology improvements. In other words, it’s not a case of introducing a chatbot; then you’re done. It’s a constant cycle of improvement. 

The increasing use of chatbots and voice interface technology is just one of 25 technology trends that I believe will transform our society. Read more about these key trends – including plenty of real-world examples – in my new book, Tech Trends in Practice: The 25 Technologies That Are Driving The 4th Industrial Revolution.