Home » Design » How to create an interesting and useful chatbot
1.00 AM How to create an interesting and useful chatbot |
How to create an interesting and useful chatbot What makes a good chatbot? Most people think about witty responses and machine learning, but at the heart of a chatbot's user experience is actually content strategy. Learn how to design a chatbot that sounds human and engages people as they quickly become ubiquitous. A bad chatbot simply repeats, “Sorry, I don’t understand” (or worse, “error”). A good chatbot feels almost human and helps answer questions so you don't have to make a phone call. or search the FAQ page. But what makes a good chatbot? What kind of table bets do people expect from a chatbot and what ruins their experience? In this article, we'll answer these questions and identify what you, as a content designer, can do to make your chatbot successful. A chatbot is a program that replicates a human conversation. Most chatbots use decision trees to generate conversation. They either recognize keywords and respond accordingly or allow the end user to select from options to guide the conversation. It is equally important to define what a chatbot is and what it is not. Let's understand what a chatbot is and what it is not. CHAT BOT IS A FORM OF CONVERSATIONAL DESIGN Chatbots replicate human conversations, and most chatbots use decision trees to do this. They either recognize keywords and respond accordingly or allow the end user to choose from options to direct the conversation. Conversational design broadly refers to any shared content, be it headlines and text on a web page, voice interfaces like Google Home and Alexa, or chatbots. So chatbot content is a type of conversational design, but it is not the same thing. A chatbot is also not a human interacting through a chat interface. This is, in particular, a computerized system. Why is it important? When design and engineering teams determine the best way to communicate with their audiences, they will most likely use shorthand. I often hear designers say, "Then we'll ask them to confirm their password." In this case, the designer may mean to "tell" the audience through text on the page, or they may mean to have a chatbot appear to inform the audience. At the early stage of concept development, it may not matter what form of conversational design the team has in mind, but ultimately the development team will be responsible for significantly more work if the end result is a chatbot. With this in mind, it's useful to clarify what form of conversational design the team has in mind. One example of conversational design without a chatbot is the conversational interface, which they design using several best practices: Headings are complete sentences. The forms have help text with specific instructions (instead of examples). The text is written in the 2nd person, addressing the audience as "you". Individual Plan Form Health insurance is designed for you. The title of one field is "Annual Household Income" and the supporting text is "Enter your pre-tax income." Another field name is "Tax Household Size" and the supporting text is "Enter Tax Household Size." The voice interface can also "speak" to the audience, and if the team believes they are either creating a competitor to Alexa, Google Home and Siri, or creating an app that these systems can download. Again, these may sound similar in concept, but the requirements are very different. Voice UI has no visual design and no ability to trigger or encourage the end user to take action. This is in stark contrast to a phone app, which can trigger notifications without the end user having to open the app first. A CHAT BOT CAN ANSWER (MANY) QUESTIONS From this description, it would seem that chatbots are the perfect answer: they can trigger notifications, they have a visual interface, and they communicate! Of course, the popularity of chatbots is partly due to these benefits. But this can create an assumption that human conversation is the best way to reach end users. Sometimes this is true - but not always! a company needs to know what its goal is and then determine whether a chatbot will help achieve that goal. This is a fantastic approach: a chatbot is a solution and should be used when it is the solution to a problem. Customer service and sales are generally good targets for chatbots. In both cases, the problem may be "how can our customer service team quickly answer common questions" or "how can our sales team help customers quickly and easily learn about a product or service without wasting significant staff time?" In these cases, a chatbot can help people get the answers they want without having to call and wait on hold. However, if the problem is “how our hospital can more accurately diagnose health problems” or “how our bank can more quickly help employees find lost wages from their employers,” a chatbot may not be suitable. A human doctor is much more accurate than a chatbot, and end users will notice. Likewise, a bank chatbot is unlikely to be able to connect to the multiple employer payroll systems needed to track paychecks. While the end user may think they need answers from a chatbot, they will quickly lose trust when the chatbot is unable to answer their questions. In short, a chatbot is not the best way to handle nuanced or extremely complex situations due to the numerous possibilities for human error. There are simply too many variables to be done "quickly" and "accurately" in these situations. Every Tuesday we send out our awesome email newsletter about UI and UX. Subscribe to get Smart Interface Design Checklists, a free PDF collection of over 150 questions you can ask yourself as you design and build just about anything, from mega dropdowns to advanced configurators. This is what it looks like. CHAT BOT IS NOT AN ALGORITHM Never forget that a chatbot is only as good as its content. Yes, the chatbot is driven by an algorithm and can be enhanced by machine learning. But before machine learning can begin, the chatbot needs a set of rules and content to speak. This is a role that should be defined by a content designer, UX writer, or content strategist We see this often in conversations about AI ethics. Voice assistants and chatbots are often called sexist and racially charged. They are not biased because the design and engineering team made a conscious choice in their favor. They are biased because they "reflect the biases in the views of the teams that created it" For those of us who create chatbots, this means that we must be consciously anti-sexist and anti-racist. We need to build chatbots thoughtfully and not incorporate machine learning until we create the content we want the AI to learn from. As with many things, what the chatbot does is only half the battle. He can “answer questions” - but what questions and how? It can "guide people to their next steps" - but what are the next steps, and how does the chatbot react when things go wrong? In other words, what will have a real impact is how the chatbot does what it does. If your team is building a chatbot, hopefully you've already done a lot of work up front. You've decided that a chatbot is the right solution. You have identified technology constraints, such as what system you will use. You explore what features will be available in this system, such as autocorrect or a built-in synonym thesaurus. Now some managers say: “Plug it in and make it work!” and you have to say, "Plug in what?!" As noted, your chatbot is not just an algorithm, and you have some content to develop. It's time to create content for your chatbot. Let's look at five best practices that will make your chatbot human: 1. DEFINE YOUR ACTIONS Since a chatbot is not a magic solution to everything, you need to focus your work on specific user flows that people can do with your chatbot. For example, let's say you're building a chatbot for a company like FedEx or USPS, you could list examples of user flows such as "track package" and "update mailing address." This means that if an end user asks the chatbot for help tracking a package, it can respond with “What is the tracking number?” But you should know your limitations. Perhaps one of the goals is to “build trust.” So if the end user says “someone committed mail fraud in my name,” the chatbot can offer condolences and quickly transfer the end user to a live customer service agent. Since the goal was to “build trust,” the team building the chatbot must be aware that anything involving sensitive information must be handled by a human, even if there are no technical or legal restrictions. There is no one right way to do this. Most organizations have some kind of value proposition or design principles that will help define the purpose of the chatbot. It is also likely that some requirements have already been defined. Therefore, the goal can be derived from a quick glance at the requirements, and the requirements become more specific once the goal is defined. without a chatbot, “our call center will receive 3 times more calls.” One might assume that their goal is to “reduce customer service hours.” This pairs nicely with their chatbot's initial request, which suggests "trending topics" it can help with - these trending topics are likely the most common reasons people call customer service. In the same interview, Banner said the bot pulls its content directly from member support content. Like many other organizations, there was a lot of useful content, but people had trouble viewing it. The chatbot asks: “Hello, how can I help you? Here are our most popular topics: STIMULUS PAYMENT INFORMATION; Checking and Savings; Mobile Banking App Guide; Home Equity and Mortgage; Automobile, personal and student loans; Digital Wallets - Try Today! » 2. SEPARATE ANSWER TYPES When you think of a chatbot, you probably think of one of two things: A chatbot that responds to everything the end user types, capturing what they want using keywords and phrases. A chatbot that follows a series of decision trees, asking the end user to choose from several options and then guiding them through a user flow. Chatbots can do one or both of these, and it's important to know what you're aiming for. In fact, even if you intend to focus on decision trees, there is a possibility that the user will go beyond the script. With this in mind, think about how you want the chatbot to respond. If someone says, “Help” or “Talk to the person,” how do you direct them? When thinking about your chatbot's word associations, remember that words have context. When an end user edits their profile and types in "phone number", they will likely want to see where to change their phone number. But if they typed something that the chatbot doesn't recognize, the chatbot says, "I don't understand," and then the end user enters "phone number" and they can search for a customer service line. This is an opportunity for developers and content strategy to collaborate to create a well-designed and well-built bot. This kind of thoughtful planning will end up in the final product. Adobe's chatbot, for example, doesn't work here. It starts by asking the end user for a random input, but after receiving a response, the bot asks the end user to select one of three options. As a user, I'm left wondering why I was asked to enter if the bot couldn't understand a simple keyword like "Adobe products." The user randomly types “What are Adobe products” and the Adobe chatbot responds: “I want to make sure I understand clearly. Which of these categories best describes your problem? Troubleshooting product problems; Explore plans and prices; Something other." Adobe chatbot asks how they can help, but doesn't recognize keywords. 3. ACCEPT YOURSELF AS A ROBOT Once you know what your chatbot can do, it's time to think about how it will do it. First of all: don't pretend that your chatbot is human. In a study with a former client, the client found that over 80% of people were comfortable interacting with a chatbot and liked the chatbot to have a name and personality. But those same people quickly lost faith in the bot and the organization when the chatbot pretended to be human. One conversation with a client revolved around whether people would talk to a chatbot if they knew they could talk to a human. Testing has shown that yes, they will! In fact, reassuring end users that a human was available (if needed) actually increased the comfort with which they could talk to the chatbot. The Hopelab team achieved similar results when they created Vivibot, a chatbot for teens with cancer. Teenagers and young adults often avoid trusting their parents or health care providers. But Hopelab has found that the chatbot removes some of the barriers. In their peer-reviewed randomized controlled trial, they were able to show that Vivibot not only provided valuable emotional support, but also reduced anxiety. “Vivibot (that's me) is a chatbot designed for young people living beyond cancer. If that doesn't sound like you, that's okay - we can still chat! Although I was created by real people, I am not a real person and do not replace the care of a therapist or other health care professional. I am not an emergency or crisis response agency. If you are injured or in a potentially life-threatening situation, call 911. One last thing: although I won't understand what you are typing, I will do my best to help you learn some new skills (and meet some of my friends along the way) . I'm over it. Let's start chatting! » Vivibot is an interesting example of a chatbot for several reasons. Firstly, the bot is not intended for one-time solutions, but rather as a tool for ongoing emotional support. This means that the bot should have a variety of responses to avoid repeating itself. Secondly, Vivibot, as a health-related bot, had to address sensitive issues. She needed to be as transparent as possible, never going into the general "sounds good" for fear of alienating the people who rely on her when they don't feel comfortable trusting people. Imagine if Vivibot seemed insensitive? Emily Cummins, a writer who wrote an article about the worst chatbot failures, gives an example in which UX Magazine's "UX Bear" asks, "How would you describe the term 'bot' to your grandmother?" Emily replied, “My grandmother is dead,” and gave her a thumbs up. This is a slightly confusing response from UX Bear, but potentially devastating for Vivibot. In the near future, we may see other states pass laws against bots pretending to be human, as California has done. While this may seem unnecessary to the world's chat bears, it is obviously important for influential or sensitive topics, be it politics or health care. 4. CREATE A TONE FOR EACH SCRIPT. When content strategists create “voice and tone,” they are two different things. Voice is like a brand's personality. It defines what the company sounds like, no matter what. However, the tone will differ depending on the situation. The voice may be “friendly,” but friendly sounds different in an error message than in a success message. The chatbot's voice should be different from the company's voice. He may say things like “oh no!” or “I’m happy for you.” when your company can't. To this end, the first step in creating a chatbot voice is to create a list of words that your chatbot says. It is important that the chatbot responds to the end user so that they know they are being heard. This means that chatbots spend a lot of time saying things like “Got it” or “I understand,” and you need to know what those agreement tokens sound like. Does your chatbot say yes, yes, or both? "Okay" or "okay"? “Excellent” or “I understand”? Options will help your chatbot sound consistent so the chatbot doesn't respond with "ok smokey" followed by "I appreciate your time," but you'll also need enough consent tokens so your chatbot doesn't sound too robotic. In the Domino chatbot, the bot alternates between agreement tokens such as “great” and “got”, but when it cannot understand the answer, it does not have an error token. The redundancy of the question "What city is this address in" (without mentioning the fact that it didn't understand my answer) initially made me think the bot had been hacked. Chatbot [Home]: Got it! Is this delivered to your home or apartment? Me: Home. Dom: Great. What is your mailing address? Me: 123 Sesame Street Home: What's the zip code there? Me: 10010. Home: What city is this address in? Me: New York. Home: What city is this address in? Me: New York. Home: What city is this address in? 5. DESIGN WITH ERRORS Chatbots, like other user interfaces, only have one chance to make a first impression. If the experience isn't smooth and easy, people won't come back. With this in mind, a chatbot should have well-written error messages. An error message from a chatbot can be as simple as “I don’t understand. Can you tell me again what you want? "But he can also do much more. For example, if your chatbot is an MVP, your error message might say something like “I can't help you with this [feature] today, but ask me again in a few weeks.” Alternatively, if the end user asks for something that the chatbot will never offer, offer an alternative such as “you can call customer service for help.” If you allow free typing, there is also a risk that someone will type a word or phrase that your chatbot doesn't understand. In this case, your chatbot may ask for clarification or even say, “I don’t understand.” But make sure you don't lock your end user in! If the chatbot can't understand after two or three tries, encourage the end user to contact a human. However, planning for failure goes far beyond just "I don't understand" or "I can't help with this." A well-designed chatbot takes into account how end users view the tasks they want to complete. Take, for example, a payroll system that can use a chatbot to help employees check their upcoming payments, tax deductions, and other requested pre-tax deductions. In such a system, the chatbot will most likely be able to answer questions such as: Perhaps the payroll system is linked to some of the employee's benefits - for example, it could have a flow built in that allows the employee to change deductions. But the payroll chatbot team needs to know that employees may approach them with related questions and concerns, such as: It is unlikely that the payroll system is also a benefits system. But the chatbot team needs to know that employees don’t think in terms of opportunities. They think in terms of needs. “I need to take care of my 401k” might mean going to one system to set up contributions and another system to change the distribution. If the chatbot says nothing in response other than “I can’t help with that,” then the chatbot has failed. Instead, our hypothetical payroll system could generate goodwill by explaining the system to the member and recommending that they speak with their HR representative. Webflow's customer service chatbot not only determines what it can do, but also proactively tells the user, “If I can't resolve the issue for you, a member of our support team will contact you. by email. Note. We do not currently provide phone or chat support as we have found it most helpful to assist you via email." You might not think of this as a bug report because it's actually about solving a problem before it ever becomes a bug. Hello! I'm a Webflow Help Desk Assistant. If I am unable to resolve the issue for you, a member of our support team will contact you via email. Note. We do not currently provide phone or chat support as we have found it most helpful to assist you via email. Internally, this means that the team must define user flows from the end user's perspective, not just from a technical perspective of what is possible. If Webflow only looked at things from their perspective, they wouldn't have thought to clarify what they weren't doing. They would simply solve problems that they could, and potentially leave users wondering why (for example) they can't find a phone number to call. Of course, a chatbot is not a person. But this is also not a second option for a person. A chatbot can help people easily get answers to their questions, help them reach out when they feel vulnerable, and simplify complex processes. This, like many other tools, is an ideal solution to many potential problems. As the creators of these chatbots, this means we have an important mission! We must create appropriate responses, humane tones, and helpful user flows. We must write content to respond to people in different moods and with different needs, anticipating their next steps and guiding them accordingly. Above all, we must create transparent and trustworthy bots so that people interacting with them can trust the information they provide. Just remember: define your actions to ensure your chatbot achieves the business goal and user needs. Create a script for your chatbot and decide whether your chatbot will respond to requests outside of the script. Accept yourself as a robot and never pretend to be human. Create a tone for each scenario. Lastly, make sure your chatbot can handle errors easily. Your chatbot is a program, not a person. However, a well-designed program can bring joy and relief to your audience! With these five steps, your chatbot can establish an almost human-like connection with your end user. Now it's your turn: follow these guidelines and let us know how your audience reacts to your bot. |
Category: Design |
Views: 54 |
Added by: admin
| Tags: |
Total comments: 0 | |