Google Design Challenge
A Google Assistant Skill for Rehearsal Room Reservation
Duration: 1 week
After interviewing a few potential end users, I decide to gather the insight from my interview and build three types of persona that represents different groups of people, reflecting their characteristic, needs and concerns, which I believe could help me further ideate my concept.
Design the reservation experience for students, professors, and community members to check availability and reserve one of the music rehearsal spaces within the new center. Provide your overall process, a wireframe flow, and one-two screens at higher fidelity.
I tried a few existing reservation system at CMU, and interviewed some potential users to get a basic understanding of music rehearsal room reservation process.
After exploratory research, I tried to evaluate the information I gathered, and created three types of personas and a journey map to help me develop my design concept.
Through Ideation, I found Google Assistant Skill would be perfect for the reservation scenario. So I utilized storyboard as a tool to visualize a story about how a user will use it across different platforms and in different situations.
For VUI, as so-called Zero UI, visual doesn’t play a significant role of it, instead, the user flow (Sample Dialog) and the logic behind it tends to be essential.
After user flow, I used DialogFlow to prototype and refine my design, and then create a partially functional Google Assistant Skill.
I conducted several user tests with my prototype, as well as role-playing, during which I got some useful insights to help me refine my design and do iteration.
Existing Reservation System
CMU has more than one system for room reservation, and they are all web-based. But the shortcomings of these web-based systems are obvious:
Too much information need to be filled, even some obvious ones, like names;
Room availability information is separated from the reservation system, users must go back and forth to finish a reservation.
Much redundant information distracts users from concentrating on their task.
For expert users(use it frequently), no short-cut, a user must go through the whole process every time. (enter, name, email address etc.)
After interviewing a few potential end users, I decided to gather the insight from my interview and build three types of persona that represent different groups of people, reflecting their characteristics, needs, and concerns, which I believe could help me further ideate my concept.
Interview & Persona
I use a journey map to help me identify users’ pain points as well as design opportunities throughout the process.
Designing a Google Assistant Skill to facilitate CMU students, staff, and professors to check and reserve music rehearsal rooms through voice interaction or Chatbot, fully leveraging the advantages of GUI and VUI simultaneously (Screen and Voice), providing a cross-platform ability (Google Assistant mobile APP, Google Home, and other wearable and in-car devices) to create a seamless reservation experience.
8 Function Modules
After classifying user scenarios, I found the music rehearsal room reservation system should centralize with 8 main function modules.
Hover on boxes to see prompts & responses
Why I choose VUI + GUI, not GUI or only VUI?
Unlike GUI’s miscellaneous process of inputting formatted information and jumping between many pages, VUI is more efficient and user-friendly, a user can say one sentence conveying a lot of useful information for the system to pick up. However, using only a VUI itself will create long-winded prompts when providing a list of information, which will give users a heavy cognitive load. In this case, GUI is still necessary.
Moreover, room reservation is a process of searching, looking up tons of information, which human beings are not good at, instead, a computer can perform extremely high efficiency when dealing with the searching task, and VUI provides a flexibility of customizing results and answers, while GUI can display these options to help users to catch up.
By comparing the shortcomings and advantages, I, therefore, choose a VUI+GUI approach for this problem.
Cross-platforms with Google Assistant
Although the reservation task best fits with GUI + VUI, a user’s whole experience also contains other tasks like “schedule inquiry”, “reminder”, “navigation”. These tasks can be achieved through either VUI or GUI+VUI. Google Assistant’s cross-platform feature (cell phones, Google Home, wearable devices, and in-car systems) allows the user to have a seamless experience among these different phases of interaction. I create storyboards to demonstrate a user’s journey through all these platforms.
Cope with errors
We can’t just design for when things work---we need to design for when things go wrong as well. This is especially true with VUIs, an error is inevitable.
But encountering errors does not necessarily mean a user must be frustrated, while by contrast, it is a critical moment to guide users to explore the capabilities and help them finish a task.
So I figure out some special occasions that could take place and I listed a response to these situations, but if I got more time, I'd love to write multiple answers to it.
Provide three carousel cards each time as options, the user can see nice images with descriptions of the rooms.
Additional prompts to build users' mental model of how could they ask.
After the reservation is finished, several Suggestion Chips are provided to guide user finish further tasks.
A user can always add details or requirements to make a more targeted search.
But without all these information, system will look up user's history records, or follow default settings to give effective recommendations.
I use the mascot Scotty Tartan as the reservation assistant profile.
for Google Assistant Skill
I did a test version of the Google Assistant Skill with DialogFlow in Google Assistant App on cellphone. I name it as Tartan Reservation ( Tartan is a CMU mascot)The video shows a basic user flow of music rehearsal room reservation.
for Google Assistant & Chatbot
I made a functional prototype with Google DialogFlow, and it is amazing that this tool doesn’t even require much coding to launch a VUI experience. The prototype is not fully functional, but it does provide a basic user flow to reserve a music rehearsal room.
Chatbot Play Guide:
Enter "Hello" or "Hi"
Type “reserve a music rehearsal room” to trigger the intent. (You could also say “rehearsal room” and so forth.
Then just follow the questions with the chatbot. (e.g, Friday 5 PM for 2 hours)
* the recommended options can't show in the chatbot, because it only works in Google Assistant.
Type sometime to try
I used my prototype and role-playing to conduct user test with 3 people, I got several insights from users, and further refine my design, especially changed a part of the flow.
The user had a hard time finding and using the wake word to activate the skill. They use“ Book”，“Order” etc. So how to make user find the skill still needs to be solved.
In my initial design, users didn’t pay much attention to the three recommendations provided, because in their mental model, they haven’t provided enough information to the system, so they directly start to provide more preference information.
Some users, especially who have no prior experience with Google Assistant, need some time to acclimate the turn-taking in the App, and they don’t know when should they talk.
For a VUI experience, there are not many efficient tools for designers to develop an interactive prototype, which is why many VUI designers and researchers leverage Wizard of Oz to do user test. But I believe a VUI prototype tool for the designer would be very useful and have a promising commercial value as well.
2.Design persona and refine prompts for the VUI
Writing prompts and responses for VUI is hard, I attempted to make the conversation as naturally as possible, simultaneously consider to keep the prompt and response concise. I believe these prompts and response eventually demonstrate a VUI’s persona as well as a product’s brand personality. Thus, how to design a persona for the VUI skill also become a challenge. I, therefore, would like to bring the project further and find the right way to design a CMU reservation assistant that is congruent with CMU’s personality.
3. Thoughts about Google Assistant App "Barge-in" Machanism
Another interesting thing I found about Google Assistant App is about the “Barge-in” mechanism. Google Assistant only allows user to speak after finishing reading the system responses and prompts, or they must press the “speak button” to barge-in which is sometimes unpleasant. Especially a novice user sometimes doesn’t know when to speak, and he/she sometimes speaks right after seeing the response on the screen, while the system is still reading out loud the prompts. Under this circumstance, it is a little bit frustrating. Since Google Assistant has a screen, so it would be nice for users to barge-in anytime when the system is asking and waiting for an answer, rather than letting a user wait until the system finishing its speech or press the button.
4. Thoughts for Google DialogFlow
DialogFlow is an AI-driven tool for developers to develop VUI or Chatbots, and it is amazingly powerful. But I wish DialogFlow could also provide more flexibility for develops and designer to customize their VUI experience. For example, DialogFlow offers Carousel Card as Google Assistant responses, but within DialogFlow, I can’t add icons, but only a hidden link. Since I want to provide the user more information to guide them to finish a task, for instance, Navigation. As far as I can find, Carousel Card does allow developers to add icons to link to Google Maps for other navigation tools. Furthermore, Carousel Card also doesn’t fit into Google Home. When I tested in Google Home, the options I provided in Google Assistant App didn’t appear in Google Home’s VUI, which is frustrated. I believe Google Assistant and DialogFlow is still an ongoing project, but it already shows its huge potential in the future.