Chatbots have come a long way since the first commercial instant messaging bot, SmarterChild, hit our AOL buddy lists in 2000. For a decade, it was the most popular artificially intelligent agent, at least when someone else didn’t have to use the landline. That was just the beginning. The novelty of SmarterChild may have worn off 20 years later, but the technology has found a greater purpose. Enter the modern customer service chatbot.
Today, we’re living in a society that thrives on instant gratification. We’re not sitting around listening to dial-up tones waiting to connect to the internet. Similarly, you’d be hard-pressed to find a person who doesn’t think it’s annoying to wait on hold just to speak to a representative. Though customer service chatbots are adept at giving quick answers to simple questions, they’re still largely underutilized. Just 23% of customer service organizations use AI chatbots, but a whopping 69% of customers prefer to use them.
It may feel intimidating to jump into the world of chatbots, but your customers will thank you in the long run, at least if you’re doing chatbots right. Thankfully, the most common chatbot mistakes are some of the easiest to avoid.
1. You didn’t make a proper plan
Building and implementing a chatbot isn’t enough. Your chatbot needs a clear purpose. That could be providing personalized brand engagement in order to foster customer retention. It could be cutting down on customer service calls, which saves you both time and money. Whatever the reason, without a solid purpose, customers won’t be inclined to use it. Create a clear strategy, then work on the functionality with the bottom line in mind.
The team at classic games site Solitaired, for example, tested a Chatbot to help with solitaire strategy. However, when users engaged with the Chatbot, none of them were interested in learning about solitaire strategy. Instead, they wanted to learn how to play solitaire. The team changed the chatbot so it could help teach the game, and subsequently saw an increase in engagement with it and the game. It’s a great example of how you have to plan and design your chatbot to match customer interest based on where they are on the customer journey.
2. Your chatbot offers vague information
Picture this: You’re at the mall, and you walk into a furniture store hoping to buy a new dresser. The salesperson who approaches you tells you about a number of dressers you can order right to your home. He shows you some different floor models, and you find one you like, but when you ask if it comes in different dimensions, the salesperson draws a blank. “But it’s made of cherry and on sale,” he says. You end up leaving empty-handed because you need a dresser with specific dimensions.
The same thing goes for chatbots. Users often use chatbots to get information about products and services before making a purchase. If your chatbot only provides vague, unrelated, or unimportant information, you stand a much greater chance of losing that sale. Instead, make sure your chatbot is equipped with details the customer actually needs like ratings and detailed product descriptions.
3. You’re making your users work too hard
Chatbots are popular because they’re easy. They’re meant to be more convenient than talking to an actual human being. You don’t ever want a customer to feel like they’re repeatedly yelling, “Can I speak to a human?” into a void that doesn’t answer. One of the most common chatbot mistakes is that the user has to work too hard to get answers.
Thankfully, this can be easily mitigated. Make sure your chatbot always moves the consumer forward. Answers should be actionable and not just drop off a dead end. Even after a conversation seems over, your chatbot should at least point to a FAQ section in case the customer has more questions.
While the experience should be easy, you should also communicate progress. The team at Unscrambled Words created a chatbot which would ask a series of questions to help their users find the right words to use in a game of Scrabble. By simply mentioning how many more questions they would ask (i.e., question 1 out of 3), they improved engagement in their chatbot by 14%. Users want to know they are working and progressing towards a solution.
4. Your chatbot doesn’t have a name or personality
Back in the early aughts, SmarterChild managed to become so popular because of its name and it’s informative, marginally sarcastic attitude. It was easily recognizable, and it felt almost human, like a friend that just happened to be created by a programmer. Your chatbot should be similar. Like all humans have names and personalities, so should your chatbot. It’s all about brand recognition.
Your chatbot’s name should be unique, easy to say and spell, and no problem to remember. You can be creative with the personality, but consistency is key. Would Siri be so popular if she wasn’t the Siri, who boldly quips that she’s reading the dictionary, and not to spoil the ending, when you ask her what’s up? It’s hard to say, but odds are we wouldn’t feel the same sort of affection for Apple’s voice-activated AI otherwise.
5. Not enough testing
You need to make sure your product is actually ready for consumers before you launch. Your chatbot needs to be adequately tested in all kinds of circumstances, both good and bad, to make sure that it operates the way it should.
Learn from Microsoft, who unveiled their Twitter chatbot, Tay, in 2016. It was meant to be a casual and conversational bot, but in less than 24 hours, users taught it how to be racist and misogynistic. The bot was quickly shuttered, but the whole situation could have probably been avoided if the worst types of language were adequately tested.
6. It’s not easy to talk to a human agent
Chatbots are great, and though a vast majority of users prefer to grab a quick answer from an AI, some people are always going to want to talk to a human. Your chatbot bot shouldn’t be the endgame of customer service. It’s just a tool that can greatly help. Make sure that if a user wants to speak to a living, breathing human agent, the option is readily available. Don’t make them work to find it, or you’ll risk losing the customer altogether.
7. You’re not making updates and improvements
Regardless of how much you test a chatbot before making it available to the public, it can still have errors or develop detrimental quirks the longer it’s in service. No human mind accounts for everything right off the bat. The most successful chatbots are regularly maintained and updated.
As time goes on, remember to regularly evaluate your chatbot and determine if you can add improvements or updates that would better serve your customers.
Just a few short decades ago, turning towards AI for customer service seemed like something out of The Jetsons. But, if AI seemed like the future, it’s because it is the future. Chatbots have given businesses the ability to help customers around the clock, even when most people are sleeping. This is a good thing, but a successful AI shouldn’t feel clunky and mechanical like the robots imagined in cartoons past. The best chatbots still have a human touch, whether it’s because of a clear, concise strategy or some smart programming along the way.
Vice President of Strategy and Marketing Services at ElevationB2B. From legacy Fortune 100 institutions to inventive start-ups, Ryan brings extensive experience with a wide range of B2B clients. He skillfully architects and manages the delivery of integrated marketing programs, and believes strongly in strategy, not just tactics, that effectively aligns sales and marketing teams within organizations.