Designing a Productivity Chatbot
Senior UX Design Project at NYU Tandon — (Bs.) Integrated Digital Media
HustleBot was born with the intent of experimenting with a productivity tool in a new media format. Prior to embarking on the project, I ran an inspiration Facebook page called Winspire, which ultimately aimed to create something more tangible and quantifiable (in terms of its impact) than merely posting motivational content.
During our senior design project at IDM, we had to call out an ambitious mission statement to steer our work; mine was to “catalyze behavioral change through new media technologies.”
Below is a high level explanation of the HustleBot prototype as it exists today, along with some learnings for chatbot design and goal setting/productivity. Though I haven’t continued building HustleBot post-graduation, the lessons learned in designing conversational interfaces as well as structuring behavioral change environments will surely guide future design work.
For a more detailed process view, see here.
So what Is HustleBot?
HustleBot is a Facebook Messenger-based chatbot designed to help users set and reach goals. By operating as a digital conversational planner, the tool employs SMART goal-setting techniques and is backed by research on the psychology of habit formation.
Through interaction, the tool helps you breakdown goals into actionable steps, and select a time to receive daily reminders and notifications to input your progress. You can also set reminders, start a pomodoro timer, listen to curated playlists (pump up tracks, study/work or inspiration), watch motivational videos and, for my own sake, submit feedback.
During the IDM showcase, one of the considerations was how to physically present a digital interface. I decided to create business cards with QR codes on the back for instant access, reducing the friction to launch the chatbot. The best part was because it’s a chatbot, I’m able to collect feedback at many points along the user journey by simply sending a nudge. Below are some lessons I learned along the way.
Lessons Learned Designing Chat Interfaces & Behavioral Change Environments
Learning was a multidimensional experience in this course in that there were many different avenues we used to gain clarity, answer questions, make discoveries and refine our work. From user testing to research materials to speaking with people in related industries, there were many forces at play to spark new insights and influence our direction.
To start, perhaps the biggest decision I made with this project was to design a chatbot to begin with…
Why chatbots?
For brands and social media pages looking to engage their existing audience, chatbots are a helpful tool. They can be designed without code, though programming knowledge definitely benefits the design by enabling more in-depth interaction design.
Chatbots are also online, can be accessed through mobile, are dialogue/texting-based (human’s natural interface) and therefore feel familiar. They’re multi-sensory, and they can even send push notifications.
This is all to say that chatbots are an effective tool — when done right.
Some things I had to consider when building the HustleBot prototype included security (ie: data sharing, access to privacy), the chatbot’s ‘personality’ (tone, pacing, language), the fact that there would be many different types of goals inputted, and my own lack of coding proficiency.
Creating digital spaces that compel positive change:
A YouTube series called the ‘fun theory’ was my initial inspiration for making HustleBot a productivity-centric chatbot. In this clever series of videos by Volkswagen, everyday objects are altered in some way to encourage use (ie: throwing trash into garbage cans rather than littering, taking the stairs instead of escalators), with the premise that if we find performing actions fun, our minds don’t perceive the work as taxing as if we found the task mundane.
Take a look at this example:
I wanted to translate this idea to something digital rather than physical, that could be entrenched within a person’s day-to-day so it helps shape behavior by performing actions repeatedly. During the research phase, I picked out a few textbooks on motivational psychology that would help garner new insights on considerations such as how to trigger certain actions for instance, or understanding how language determines the likelihood of receptivity.
The books I read were:
1. ‘Social Media and Your Brain: Web-Based Communication is Changing How We Think and Express Ourselves’ a collection of essay by many authors, published by C.G. Prado.
2. ‘No Sense of Place: The Impact of Electronic Media on Social Behavior’ by Joshua Meyrowitz
3. ‘Designing for Behavior Change: Applied Psychology and Behavioral Economics’ by Stephen Wendell
To attempt to distill all of the information these books offer would do them a disservice, so I’ve selected a few samples of the most impactful quotes for my design work below:
1) “If the claim that habits make us who we are seems too strong, consider one basic and obvious way that habitual behavior determines personality or personhood: by establishing what can be described as responsive continuity. This continuity, or behavioral consistency… is a similarity of responses or reactions that constitute what others see as an individual’s… character. In a straightforward sense, habitual behavior determines what is normal for individuals to do most of the time, and therefore also what is likely they will do in unusual or exceptional cases. It is in this way that habits mold people, and it is in this way that habits change people.”
~C.G. Prado
What this quote suggests is that habits are linked tightly with an individual’s sense of identity. To attempt to change a habit then is very much a gradual reformation of one’s self-perception, whether that change occurs consciously or not.
One way this can be achieved is by carefully selecting the language of the dialogue to frame how users interact with the chatbot and guide them towards more positive self dialogue. Aha! If you can figure out the best language to shift a person’s perception of their own capacities, you can get them to perform tasks that reinforce those perceptions.
Structuring spaces that foster behavioral change should therefore be dynamic, changing over time to accommodate an individual’s progress.
2) “Even our language about behavior change can mislead. For example, you don’t ‘break’ a habit. That’s the wrong verb. It implies you exert sudden force and the habit goes away. A better verb would be ‘untangle,’ because it sets the right expectations of how to get rid of such behaviors. It requires persistence.”
~Stephen Wendell
Diction as a tool is incredibly powerful because it frames expectations. You can more accurately understand what action you’re required to perform, how to perform it, whether its sensible, and what the result will be based on how its written.
For HustleBot, this means speaking as ‘human’ as possible (I coined the acronym AHAP) while adhering to principles of behavioral change. More generally, it translates to developing a personality that aligns with how a user would expect someone on the other end to act based on the context of their conversation and the industry that they’re in.
This can manifest in everything from how long the chatbot takes to type out sentences, to their use of emojis/gifs/other non-text media, to the literal words a chatbot uses to describe things and hold conversations. Even the way a chatbot ‘reaches out’ (ie: push notifications, cross-platform messages, etc…) should feel non-invasive and AHAP. The concept of how we say something, not just what we say, is therefore incredibly important — especially so in an online, text-based format.
3) “How have digital technology and social life become so entangled? Part of the answer is that attention, emotion, and desire have unique social functions in the age of social media. Social media operates according to an attention economy. Apps like Facebook, Twitter, Instagram, YouTube, Snapchat, and others revolve around built-in metrics of attention — likes, retweets, followers, and so on — so that using social media apps is like playing a game of vying for attention points. As our scores get higher, our desire attention points increases. When we value another person’s opinion or want to express that we care about them, we pay them in attention points. Many of us now have unquenchable thirst for measurable quantities of attention, but at the same time yearn for social goods that are harder to measure, like authentic relationships. The attention economy has conferred such high value on the very idea of social life, it is now both a commodity and a symbol of pricelessness.”
~ Joshua Meyrowitz
Are there shareable elements a chatbot can include? Can we rally someone’s support network behind their goals to increase the likelihood that they stick to them?
Future plans beyond the scope of the class are to expand the text-based interface to voice technologies such as Alexa or Google Home. This would of course entirely reimagine HustleBot’s interaction based on the differences between speech vs. text based interfaces, but it would also expand it’s accessibility and practicality.
Since the original point of HustleBot isn’t the chatbot itself per se, but the effort to help people commit to their goals, its important to be adaptable and figure out what really is the best media format to do so. Should there be something that proves better, I wouldn’t hesitate to make the switch.
