We communicate through images, movement, sounds, text, symbols, objects…
Bots are evolving so fast! Last October I wrote about how bots are the new apps and shared my thoughts about why bots and about the opportunities they create. Over the past few months, interest has grown and we as a community have iterated on our approaches to build bots. For some, the hype has gotten to a point where everyone wants one, but do we really understand them or the efforts that are required to build them.
So far, my initial thoughts have stood the test of time.
The following are observations, experiences and thoughts, about what we need to consider when we set out to build a successful bot.
Bots are the new Apps – Part 2
Bots are defined by an exceptional user experience! And not by the amount of Artificial Intelligence (AI) or by Natural Language Processing (NLP) that is used to build them.
A bot is successful if users actually use it! This is important, because we must build our bots with meaningful telemetry, logging and data collection mechanisms that empower us to measure, validate and iterate. This data forms a foundation that we can use to hypothesize and prioritize our efforts. Then through A/B testing we confirm that we solve the user’s need in the quickest and easiest way possible. This in itself directly impacts our business model and pushes us to grow the end-user consumption and shift our business model to selling through microtransactions. In other words, a bot helps users be successful.
A bot has a clear purpose. Focus every design decision into driving that goal.
A bot must be refined over iterations. It’s almost impossible to get them right the first time. Iterate, test, fail, learn and try again. When analyzing conversations and interactions, don’t focus on a single message. Instead analyze the entire thread to extract its full context. Then review the users’ interaction history to find trends and identify new opportunities.
Users love buttons!
Users want to get things done quick and easy. Feed this need by guiding user interactions with UI controls (Cards, Buttons …), then by text and finally by voice. Try to keep voice for scenarios where other options aren’t appropriate. (e.g. while driving) Furthermore, a bot is just another type of User Interface (UI), and it can be limited in amount of context that it can communicate at one time. So if a standard channel (SMS, Email, Twitter, Messenger, Skype, Slack, Teams…) can’t provide the right experience, consider building more than one type of UI. Think about integrating the bot in Apps and Websites.
The user experience is everything!
Focus on the actual user experience. If it doesn’t feel right, it probably isn’t! Try not to abuse Natural Language Processing (NLP). Leveraging only text or just voice can lead to unnatural user experiences. As humans, we communicate through images, movement, sounds, text, symbols and objects… It’s important to keep in mind, that users are often underwhelmed by bots that have not had enough though (a.k.a. tender love and care) throughout their conception. Let’s not kid ourselves, building a bot requires considerable amounts of design. Be sure to storyboard/spec out dialogs and flows. And most importantly, be certain to have a clear exit criteria/goal for each user interaction.
Fail gracefully. Leverage phrases like “I’m sorry I’m not following you” or “could you rephrase that, I don’t understand” and “let me try that again” to engage the user to provide more context. Users rarely express themselves the way we expect them to. Take this an opportunity to learn from each user interaction.
Bots are intelligent Apps with domain specific knowledge
Be responsive and inform the user about what is going on. Use phrases like “let me check on that,” “this will only take a second,” “bear with me for a moment,” or “hold on. This is taking more time than I expected” to provide users with perceived performance. This strategy plays a crucial role in keeping users engaged. A user who is left waiting, perceives the delay as an inefficiency, an error or a limitation in capabilities.
Do keep track of what has been communicated and by whom. This level of scrutiny allows developers to adapt the bot to a user’s identity, preferences, regional dialect and language level. Above all else, be sure to use and adapt the proper level of formality (a.k.a. etiquette). If we fail to use the right language, we fail to engage and communicate with the user.
Identifying paths in the conversations, learning and continuously improving models requires a significant amount of data. Consequently, we must be responsible and take personally identifiable information (PII) seriously.
Personally identifiable information, or sensitive personal information, as used in information security and privacy laws, is information that can be used on its own or with other information to identify, contact, or locate a single person, or to identify an individual in context. The abbreviation PII is widely accepted in the U.S. context, but the phrase it abbreviates has four common variants based on personal identifying. Not all are equivalent, and for legal purposes the effective definitions vary depending on the jurisdiction and the purposes for which the term is being used.
Failing to inform and gain consent from users, is a direct violation of trust and in many jurisdictions can lead to legal action. Loyalty is built on trust, and this comes with lots of strings attached. Be sure to familiarize yourselves with the effective privacy legislation that is applicable in your markets’ geopolitical boundary.
A bot empowers the user to achieve more
Be creative and natural in your approach. Leverage channels, languages, images and UI Controls to help the user discover what the bot can do. Keep it simple and focused. Ask questions, challenge and observe the way your bot is used. Leverage insights gained from data analysis and test your ideas. Remember, it’s next to impossible to get it right the first time, don’t rush it.
How can bots help you get things done in the quickest and easiest way possible?