The history of chatbots dates back all the way to 1966, when MIT professor Joseph Weizenbaum made ELIZA, which took words from users and paired them to a list of possible scripted responses. ELIZA was supposed to mimic a psychotherapist, yet users were sharing their deepest thoughts with the bot.
That trend of users regularly offering up information to chatbots continues today. Add in voice assistants, which have become a ubiquitous part of the home, and we’re spending plenty of time every day with bots.
While the aim of chatbots is to make life easier for users, sometimes the line gets blurred between bot and human. That can lead to uneasy user experiences.
Susie Fawzi, a consultant specializing in customer experience, recalls the time she encountered a chatbot for a luxury brand.
“It was supposed to look like a person was on camera chatting with you,” she says. “You could tell it was not a real person, but it was close enough to be uncomfortable.”
That added element of personification can be helpful in certain situations, but it’s a fine line. Make your chatbot too impersonal, and visitors won’t feel welcome. Make it too close to a real person, and you just might scare off your customers.
“Frankly, it was creepy,” Fawzi says. “The bot was making facial expressions and blinking, but had this kind of image of something from the movie AI. At first, I was wondering, ‘Is this real?’”
How, then, can brands use chatbots in a way that’s helpful to customers, but doesn’t leave them thinking like they’re about to enter a dystopian sci-fi film?
Don’t try to fool your customers…
The best way to prevent a chatbot from feeling too human? Don’t pretend it is one. Be upfront that the conversation is between customer and bot.
We’ve all encountered an online conversation where we weren’t entirely sure whether the other party was a live agent or an automated system. And while there are ways to confirm who you’re talking to—such as asking what color shoes they have on—there’s already a sense of distrust.
Even if a customer can infer early on that they’re speaking with a chatbot, there’s little to gain from trying to deceive them. And you may find chatbots can be advantageous on more sensitive subjects.
“Users will be more forgiving when they know they’re dealing with a bot,” says Stefan Koujoharov, founder of Chatbots Life. “They’re also willing to share more personal information because they don’t think a bot will judge them.”
Chatbots are meant to help solve customer problems. Once it’s identified itself as such, the chatbot can either start finding a solution or pass the customer along to a live agent if they prefer to speak with one. Some users may be fine interacting with bots, but an audience made up of Boomers will likely prefer chatting with a live agent.
…but consider their preferences
A bot’s behavior should depend on several factors. You’ll want to consider your customers’ age and their ultimate goal in using the chatbot. Will customers be placing orders through it? Finding information? Taking surveys? Identifying what’s most helpful goes a long way in an effective bot.
“Where personality and focus factor in is in the details, such as bringing a bot to life through its personality, which can encompass its name and how formal, informal, or humorous it is when responding to questions,” says Michael Ringman, CIO of TELUS International, digital solutions provider. “If a customer is making financial transactions online, do they really want a bot to have a quirky tone and use emojis as part of the process? Likely not.”
Additionally, if your brand is global, Ringman notes it’s worth working with local experts in different countries to give your chatbot a voice that speaks to people across the globe. Diversity and inclusion efforts shouldn’t stop at the machine level.
Add some personality
The fact that a chatbot is a machine doesn’t mean you should avoid infusing it with any personality.
“Given the frequency with which today’s consumers deal with smart speakers and virtual assistants, most will likely respond more favorably to a chatbot that’s not boring,” Ringman says. “Your chatbot might be a machine, but the ultimate goal is always to make an authentic connection.”
It’s likely your users will assign a personality to the bot anyway, and that assessment may change based on the tone of its responses.
Design firm IDEO ran an example testing this theory with an SMS chatbot. It only sent text messages to users, encouraging them to exercise based on data from their smartwatches. When the bot was more pushy and demanding, users referred to it as “he.”
“You have your logo, brand voice, and style guide and how they fit in with your company values,” Koujoharov says. “That’s no different from how you’d approach a bot. You may as well design a personality for it and use that to trigger the right thoughts and feelings about your product.”
Continue to tweak and adjust
Like any marketing endeavor, chatbots aren’t something you simply set up and let run on autopilot. They should constantly be monitored for performance and adjusted as you learn more about how your customers interact with them.
You may find that a majority of your customers prefer a chatbot with some personality, but they don’t like when it’s cracking jokes or confusing them about the task at hand.
Ringman notes that usage analytics, follow-up surveys, and verbal feedback to live agents must all be considered when making changes to a chatbot. They’re a direct line from your customers, so take advantage of them.
“Consumer feedback should be updated regularly to expand its scope as needed in order to increase the frequency of usage that will free up additional live agent time and attention to focus on more complex issues,” Ringman says.
The goal of gathering this chatbot feedback is to increase customer satisfaction. Brands can find out key information like what people are understanding about their service, what customers are struggling with, and where they’re dropping off. As Koujoharov notes, when you’re armed with these insights, you can make your entire sales funnel more effective.
Here are five examples of chatbots that work well and why they’re so successful for their customers.
Amtrak introduces customers to Julie
An easy way to provide customers a human feel without going overboard? Give your chatbot a name. Amtrak went with Julie, and she’s been a hit with customers.
Many people contacting Amtrak are either making a reservation or asking about schedules, fares, or train status. Julie provides a way to answer those questions in a relaxed, stress-free environment. She recognizes more than 46,000 city names and can help with most questions guests have. For ones she doesn’t know the answers to, Amtrak provides a “Contact Us” button to chat with a live person.
On average, Julie handles 50,000 calls per day, though she may get up to 95,000 during peak travel times. She also takes in more calls in one day than a human agent does in an entire year!
Capital One keeps you on top of your finances
Have you ever forgotten to pay a bill? Or signed up for a free trial with the plan to cancel before you had to start paying, only to have it slip your mind, resulting in a charge you didn’t want?
Capital One’s chatbot, the cleverly named Eno (“one” backwards), wants to make sure that never happens again.
Eno will chat on whatever platform you prefer—phone, browser, messages, inbox, or smart watch. It can provide answers and virtual card numbers at checkout, but Eno truly goes above and beyond with its monitoring capabilities.
Eno tracks users’ accounts and sends an alert when something is unusual. For example, if you made a significant purchase at a store you don’t normally shop at, or have a higher than normal bill with a utility company, Eno will check in to make sure everything is alright.
The chatbot can also predict upcoming charges. When you sign up for a free trial, it’ll ask if you want a heads up the day before it expires. That way, you can stay on top of upcoming subscriptions and can adjust your financial plans accordingly.
And because Capital One knows its customers are constantly on the go, it allows users to reply with a simple emoji (say, a thumbs up) to confirm an action.
In many cases, it’ll save users the hassle of having to call a customer support line. And that’s more time they can spend on other activities.
Wingstop knows people are hungry on social media
Visit Wingstop on Facebook and a chatbot will pop up with a simple message: “Heard you had a case of the Crave. We can fix that.”
Hungry customers can simply type in “Order” and Wingstop will pull their nearest locations. The chatbot displays full menu items and prices so customers can easily select your wings, sauces, sides and drinks. In a few moments, they’ll have wings and fries ready to go.
The chatbot is also available on Twitter, and it plays into the simple idea that people browsing social media are hungry. It’s not a far-fetched thought: Instagram hashtags like #foodporn are prevalent all over the platform, and spotting a well-placed photo on a feed can cause hunger pangs that need to be immediately satiated.
Wingstop has chatbots on Alexa and text, too. Customers text “Order” to WINGS (94647) and a multiple-choice menu will pop up. Registered Wingstop.com users can pay with a credit card on file or pull up a previous order.
While some people may head over to Wingstop’s website on their phone or laptop, its chatbots cut out those additional steps. Because when you’re hungry, the less you have to do to make food appear, the better.
Absolut Vodka offers a freebie
Something people like more than supporting a brand they love? Getting something for free from a brand they love.
Absolut Vodka played into that emotion with its seasonal Facebook Messenger chatbot, which offered a free drink to anyone using the bot (provided they were over 21, of course).
Users could find bars nearby that featured Absolut through the chatbot, which pointed them in the right direction and provided them with a code. Once at the bar, the user would give the bartender the code and could enjoy a cocktail.
But Absolut took things to another level. Once the bartender served the free drink, the Absolut chatbot would follow up with an offer for a Lyft ride to return the user safely home.
By providing a memorable experience, Absolut was banking on users returning as customers, picking up a bottle the next time they were hosting a party at home.
They also did a great job collaborating with Lyft to provide transportation. Finding relevant ways to integrate with other brands can be a major differentiator in the chatbot experience.
Slackbot keeps internal communications flowing
Koujoharov notes that internal chatbots are more popular than external ones, and they can help give business operations a boost. Slackbots, the robot assistants of the messaging tool Slack, are the gold standard for office bots.
“Slackbot is a great example of how an internal bot can improve company efficiency,” he says. “That’s where they really took off first.”
In addition to being helpful around the workplace, Slackbots are effective because of how simple they are to activate. You don’t need to know any coding, and you can set one up in just a few minutes.
Slack has three main kind of chatbots:
- Reply bots, which trigger when someone uses a certain phrase. They can be helpful in answering questions or topics that frequently come up.
- Notification bots, which monitor information from other apps and share them with you and your team. For example, you can integrate a Twitter or blog feed that will notify relevant people when a new post is made, allowing you to quickly reply.
- Slash command bots, which trigger an action when you type in a command. They may call up a third-party app to help perform the task, which can be helpful, like setting up a reminder or an office poll. They can also trigger sillier tasks, like pulling up the perfect gif from Giphy.
The bots all work in tandem to make collaboration among teams simple and more effective. And with the ability to add information to the chatbots, Slackbot is constantly growing and improving.
Image via Pixabay