Chatbots are an increasing part of our daily lives, redefining how we engage with the internet and with businesses. Canadian messaging company Kik explains it like this: “First there were websites, then there were apps. Now, there are bots.” Just like the early internet, bots are set to transform commerce as we know it, making it easier than ever for consumers to reach, engage, and transact through instant commands.
Wikipedia defines chatbots as computer programs that conduct conversations via auditory or textual methods. Today’s marketers see bots as the next big thing — a way to get instantly closer to customers. Chatbots create a new level of intimacy between consumers and brands with instant accessibility, always-on service, and superhuman abilities. They are popping up everywhere from our computer screens to our phones to our kitchen counters.
As today’s companies face an increasingly impossible demand for perfect service 24 hours a day, 7 days a week, chatbots become even more compelling. Most companies agree that chatbots are a new “must” — but the next question becomes: How? How can companies leverage chatbot technology to enhance their relationship with customers? Answer: Don’t just make a chatbot. Make a dynamic chatbot that is an extension of your brand.
Chatbots raise some fascinating branding questions. What does the voice of your brand sound like? Is it male or female? If your brand took the Myers-Briggs test, what result would it have? Is it smart? Funny? Serious? How do you want your customers to feel after they engage with your chatbot?
Everyone knows that bots are smart. They can outperform humans in terms of speed and sheer access to information. But what if they also had emotional intelligence?
That’s the question I asked this summer when I met Jurgen Mantzke, a pioneer in the new field of UX design for chatbots, at a marketing event in Seattle. Mantzke believes that EQ, or emotional intelligence, is just as important as IQ when creating today’s conversational technology. He coined the term “machine empathy” as a key ingredient to bot design. At Microsoft, he helped design Cora, an internal chatbot that assists with executive sales. I was so fascinated with his take on UX design for chatbots that I couldn’t wait to share the five biggest things he’s learned about creating effective bots that facilitate strong human connections.
1. Accommodate the user, not the chatbot
Many chatbots force users into stock phrases, making users feel limited and constrained to unnatural phrasing. This is legacy behavior from early command line interface, which used shorthand codes. If a human needs to accommodate a chatbot, then it’s a UX fail. Today’s users need to feel the freedom of natural word selection and trust that the chatbot will understand.
Improvements in natural language processing happen every day to help make chatbots more dialect aware. Linguists focus on localization that will help chatbots recognize the smallest inflections. Chatbots are also able to learn from every encounter so they get better and better at servicing individual users.
2. Create conversational flow diagrams
Try to avoid approaching chatbot engagements as simple Q&A sessions. Include a chatbot UX designer in the design process so they can diagram and map out actual conversations and dialogue. You should test these diagrams with real customers to ensure you are in touch with consumer needs rather than projecting biases. Think of your chatbot as a modern concierge for your customers, eager to service their needs; don’t lock them into rigid conversations. Modern chatbots should be able to accommodate everything from “Order me the same pizza I had last week” to “How far am I from your nearest store?”
To create a more natural flow, it is also important to feed your chatbot a variety of responses (upwards of 20 for each scenario) so the chatbot can piece together unique phrases instead of repeating itself over and over. Talking like a broken record can quickly frustrate users.
3. Keep it quick
Like any good concierge, chatbots should require the user to do the least amount of work possible. Companies should use yes/no boolean responses to keep the conversation moving. It also helps to provide users with quick reply buttons and allow them to toggle between responses. Companies should also allow their chatbot to complete forms by remembering pertinent data and information — no one wants to manually fill out their address and phone number every time. Remember, quick and efficient encounters make for happier users.
4. Maintain context in conversations
Although maintaining context comes quite easily to humans, it is still one of the hardest things for a chatbot to do. For instance, when you quickly follow up the question “What is your favorite restaurant?” with “Where is it?,” chatbots can quickly get lost. The more context your chatbot can glean, the better. More context means more machine empathy, which drives better engagement.
Think of chatbot encounters like snowflakes: No two encounters should be exactly the same. Chatbots should be able to leverage dynamic context to create personalized responses. Key dates, weather, schedules, etc. are all important factors when creating responses. The more intimately a chatbot knows a user — what car they drive, when they get up, what they eat — the better the chatbot can perform. All companies should include permissions within their chatbot technology to protect user privacy.
5. Inject personality
Refer to the Myers-Briggs personality types (MBTI) as a baseline for choosing your chatbot’s personality. For Microsoft’s chatbot Cora, Mantzke chose a proactive, assertive personality type to best serve busy sales teams. Myers-Briggs helped him and engineers humanize Cora and deliver consistent, lifelike responses.
Certain verticals lend themselves to certain personalities, but all companies should use their chatbot as a unique branding opportunity to differentiate themselves and delight users. For example, Microsoft’s personal digital assistant Cortana got her name from the main character of the Microsoft video game Halo. Microsoft used the same voice in the game and the chatbot. And consider Poncho, the popular weather chatbot, which offers punchy daily weather forecasts delivered by a cat. He’s also equipped to talk on a wide range of topics, which works well, as he can quickly pull the topic back to the weather.
Poncho, the weather chatbot launched in 2014
Whatever business goal(s) your chatbot serves, it is important to remember that the entity on the other side is human. While a sterile, robotic chatbot can be frustrating, a great chatbot can create stronger connections with your customers and build brand loyalty. In China, almost 1 billion people now use WeChat on a daily basis as a way to work, order train tickets, book dinner, play the lottery, even do online banking as they move towards a cashless society. Instant messaging is no longer about messaging — it’s about instant everything. Your customers are ready to experience your brand in this new, powerful way.
Christi Olson is the head of evangelism for Bing.