When we walk into a fancy hotel, we tend to look at the representative in Reception Counter that is if they have not already greeted us with a smile. Because it’s what they are supposed to do. When someone visited their hotel, they greet the visitor and offer to help them with what they want. That’s just good Customer Service. We probably ask them- how much does it cost to stay there for a night? what are the facilities available in the suite? Is there a free Wi-Fi? Will there be complimentary Breakfast and so on. They patiently address our queries and provide answers after which we make a decision to stay in that hotel (although that’s not the case anymore since the advent of World Wide Web!) Such is the case of a Bot.
When you visit certain Websites, you might see a pop up that greets you “Hi <Your Name>” or the chat bot may appear when you click for help (Example: Citibank). Although this may not be the only purpose of bots, these Chat Bots are designed in such a way that they stimulate conversation with humans over internet about the services offered by the website and more. Chat Bots follows the basics of contextual communication.
Every interaction with a human user should be as meaningful as possible. Otherwise, Bots do not add much value to your service. So what are some of the things to keep in mind when developing a bot? Bitcot tells you how it’s done the smartest way!
Choose your Text Elements carefully
When you are creating a bot, keep in mind to add options (in the form of buttons, for example) to help restrict the choice of user to the answer that you have programmed to give them. If your choice answer is running text, it opens up a lot of possibilities which makes providing a “Definite” answer difficult.
In cases where answers could not be limited to options, try programming it in such a way your response (the Bot’s response) comes with a question that has “yes” or “no” or “I don’t know” options. Some of the platforms that let developers build, host and deploy bots are Chatscripts, Pandorabots, Imperson. And make sure the platform you pick to build your bot caters to addition of buttons. Of course using Buttons make the design complex. In such cases, it is better to indicate options to users to respond.
Which part of the world are you from? (Please indicate one option only -AM for America; AS for Asia; EU for Europe; AF for Africa and AU for Australia)
Teach Language to the Bot
Ever since the advent of the concept of Text Messaging, the usage of one language has varied extensively. How often have you encountered everyone in your contact list asking “How are you?” the right way. The possible ways one can ask could be “Howa you?” / “Hw r u?” / “Hw r you?” / “Hw are u?” / “Hw are you?” / “Hwa u?” and so on…
Imagine the possible combinations of every common question that you have to account for when coding for a Bot. That’s how intuitive your tool should be. Building a chatbot with AI behind it could be one of the hardest things in your life.
- Front load your user base (given you have one) with people that like breaking things to test your system. Onboarding the right people up front is crucial.
- Use an SEO typo generatorto plan for common misspellings and errors.
- Accept that building your AI is going to take time and that initially it’s not going to work well. The only way it gets better is when it breaks. Plan for this and create boundaries to limit breaks.
Make sure Boundaries are accounted for in your design
For any apps, regardless of the design you always specify limits such as Screen Resolution or Width and so on. So why should a Bot design be any different? Create boundaries so that you eliminate some of the ways in which user might break your Bot. No matter how awesome and intuitive your design is, your Bot is always going to break at some point. So its better to take a bow in the most elegant way as possible.
When a user asks “How’s your day?” Your Bot can respond with a “So far so good” standard response or take control of the conversation by steering “Better, I need your Time zone to get started. Where are you from?” Or any better response for that case. The point is, don’t ever leave a response open ended.
With these design considerations in mind let’s check out Zikabot that uses conversation as a prototyping tool to test all assumptions now that SMS is evolving as an interaction model for creating new services and user experiences.
Case Study: Zikabot
Zikabot was a service created in Puerto Rico engagement channels online (like SMS) where people can anonymously ask different questions about Zika Virus and the bot gives them accurate health information. The Bot was a prototype but in the backend a lot of Subject Matter Experts took matters in hand (considering that the response impacts a lot of lives) and answered inquiries that required manual intervention.
- When having a Bot conversation with Users, Subject Matter experts observed that the tone was very essential to actually engage users in the conversation.
- For example, since bot conversations are anonymous, this enabled human users to actually ask some questions that were personal like “Why is the Zika virus in Puerto Rico?” “How is it possible that the same mosquito can spread three different viruses?” “Does Zika affect pets?” “I’ve had a lot of intercourse lately—what should I do?”
- When designing a response, you have really got to keep in mind that in cases like this, Zikabot, where people’s view and beliefs (in turn their life) depends on your response since this is regarding creating awareness. The more responsible you design responses, the better is the user experience.
- SMS can be a safe space for tough topics. Have a clear objective, especially with sensitive topics.
- Start with Questions.
- Establish a set of Parameters.
- Iterate in real time
- Build Boundaries.