The hype around chatbots has been huge.
In 2016, Matt Schlicht, CEO of Octane AI, stated that bots will eventually kill apps and websites.
In 2018, Gartner predicted that by 2020, 85 percent of engagements with businesses will be done without interacting with another human — and that 25 percent of customer service operations will be handled by virtual customer assistants.
There’s a lot to like about chatbots. They offer brands stores the opportunity to provide 24-hour support to shoppers, reduce customer service expenses and even increase sales.
But there’s a big problem. Chatbots, for the most part, aren’t cool, and the majority of customers don’t like them.
PointSource’s 2018 Artificial Intelligence and Chatbot Report notes that 59 percent of users get frustrated if chatbots don’t offer a path to resolution within five minutes. Fifty-one percent of users don’t believe chatbots understand what they are looking for.
CGS’s 2018 Global Customer Service Survey was equally damning of chatbots. That survey found only around half of respondents would choose a chatbot to solve a quick customer service query.
Building a chatbot that adds to the customer experience rather than detracting from it is hard. But that doesn’t mean it’s impossible. By understanding the underlying problem with today’s chatbots, brands stores can design their virtual assistant to break the mold.
What’s the Problem With Chatbots?
The huge hype around chatbots is part of the reason so many are awkward to use. Intercom VP of Product Paul Adams thinks many companies are blindly applying chatbots to solve any customer problem, even if they have a better and fully functional solution already in place. It’s not just a case of bad design or bad programming — though they certainly play a role. Rather, bots are being used to solve the wrong problems.
While automated systems can be programmed to handle typical scenarios, they cannot adapt themselves to exceptional circumstances, says Daniel Polani, a professor of AI at the University of Hertfordshire in the UK. The trouble is that some situations don’t just require human problem solving; they require a level of empathy and compassion that chatbots simply can’t provide.
Chatbots may never be able to understand empathy, but they should be able to hold a conversation, right? Unfortunately, this often isn’t the case.
Marketer Abhishek Anand explains that making a chatbot that “understands the depth and context of a conversation” is challenging. Asking a bot 100 different questions isn’t the same as having a conversation. Instead, the bot is having 100 different micro-conversations.
Natural language processing (NLP) is the part of artificial intelligence that focuses on helping computers process and understand human speech and text. While big improvements to NLP have been made, the vast majority of bots are still scripted, glorified IVR systems, says Amit Paka, founder of Fiddler Labs. “Sway too far from the script and you get the dreaded ‘I don’t understand the question,’” Paka says.
This can lead to massive failures in the user experience. When chatbots fail, they tend to do so spectacularly, DigitalGenius’ Juan Ageitos says. Rather than providing assistance, customers can get stuck in a loop where the chatbot asks the same question over and over again — until the user gets frustrated and quits.
How to Make You Chatbot Useful and Not Awkward
The issues with today’s chatbots make it hard to create a tool that customers will enjoy using. But socially graceful chatbots do exist, and you can have one too by following the advice below.
According to the UX Collective, the best chatbots out there make it clear from the start of the conversation that users are chatting with a bot, not a human. Knowing that they are talking to a bot gives users more realistic expectations about what they can achieve. That should also make them more forgiving when the conversation doesn’t go quite right.
In other words, don’t try to trick your customers into thinking they are chatting with a human. It’s only going to end badly.
In fact, it’s precisely when customers hit a dead end or get stuck in a loop — and it dawns on them that they are talking to a bot — that the experience gets negatively perceived, says Re:amaze cofounder David Feng. When expectations are set correctly, customers usually aren’t put off by a chatbot’s limitations.
Your bot needs to own failure, too. When it does fail, it is important that the bot do so with humility, says designer Micah Bowers. Chatbots must acknowledge that confusion exists, assume the responsibility, and provide options for moving forward. Bonus points if one of these options is to speak to a human.
Your chatbot isn’t just a customer service tool. It is an extension of your brand and should be considered the face of your online store. As such, it needs to act as a brand ambassador.
This means your bot should have knowledge of who your company is, what it stands for and how it works. A chatbot that doesn’t have this knowledge is a serious liability, says Mariya Yao, CTO and head of product at Metaraven.
Yao recommends going further. Give chatbots access to the majority of your existing systems — customer databases, marketing kits, brand assets and CRMs — so the bots can get a deep understanding of your customers, she says. This, in turn, will facilitate greater personalization and more efficient service.
Give Your Bot Focus
Your chatbot shouldn’t do everything, and your customers don’t want it to. In Drift’s 2018 State of Chatbots Report, 35 percent of respondents said they would use a chatbot to resolve a complaint or get detailed answers and explanations to questions. Only 13 percent would use it to buy an expensive item.
If you don’t know what you want your chatbot to achieve or what process you want it to help with, you should put a hold on development and figure this out first. Your bot needs to have clear goals, says EventAgent founder Nikhil Vimal. Rather than ride the hype train, make sure your bot does something to speed up a known process.
The most successful eCommerce chatbots have this kind of focus. Sephora’s chatbot on Kik is focused on providing customers with makeup tips and product recommendations. H&M’s bot asks questions about a user’s style and creates an outfit for them to explore. American Eagle Outfitters’ Aerie app helps users find the right product by using a “this” or “that” option to narrow down tastes.
Give Your Bot a Personality
Just because your users are talking to a bot doesn’t mean the conversation has to be robotic.
“Personality is what makes conversing with a bot impactful, relevant, and engaging enough to forge a lasting connection between brand and customer,” says Chatkit’s Andrea DeLong. Without a defined personality, your chatbot is almost certain to be forgotten.
Janis.ai co-founder Josh Barkin recommends getting creative with fallback messages. “[Neither] you nor your users want to hit a dead end, but if your fallback responses are creative enough, then you might just delight users with the unexpected.”
The Manifest’s Rhonda Bradley recommends having your chatbot mirror the personality of your buyer persona. Have your chatbot reflect the speech and mannerisms of your typical customer. This will help put users at ease, create a more enjoyable experience and ensure that your chatbot folds seamlessly into your overarching brand strategy.
Your chatbot isn’t going to replace your customer service department or transform the way customers shop online. But your bot can add to the customer experience. Keep it authentic, give it focus, integrate it into your business and add your brand’s personality to create a chatbot that is much cooler and way more useful than those of your competitors.