Design Thinking for Data Scientists — Brainstorming, Solution Design, and Prototyping

DSXL4AIML
8 min readSep 27, 2020

Author(s): Tara Su

Abstract. This report documents the brainstorming results of the proposed improvement of the iPhone app “mailboxes”. After the brainstorming, 3 ideas on the change of current interface, virtual reality (VR) solution, and voice command were selected. Paper prototype, textual prototype, and wizard of oz prototype were used to represent the solutions. The tradeoff and suitability to users were analyzed accordingly.

Brainstorming plan

Before brainstorming, I first decided on the below plan:

  1. Have 3 individual brainstorming sessions across 3 days. Each session should be between 1–15 minutes.
  2. Come up to at least 20 ideas.
  3. Put the function requirement at the top of the scratch paper.
  4. Come up with ideas not related to the interface.
  5. Come up with ideas that use untraditional technology.
  6. Think about solving the problem more broadly. Don’t limit ideas to improve the interface. Think about how to achieve the ultimate goal.
  7. Think about the users that are out of scope.
  8. Think about the scenarios that are out of scope.
  9. Get crazy and have fun!

Brainstorming execution

Figure 1: Brainstorming notes. Left: day 1; middle: day 2; right: day 3.

Figure 1: Brainstorming notes. Left: day 1; middle: day 2; right: day 3.

During the three short brainstorming sessions across 3 days, I wrote down the notes shown in Figure 1.

Selection criteria

The selection criteria I used were:

  1. Have functionalities that address one or all 6 requirements of 1) sort emails into customized folders 2) making notes while reading emails 3) adding calendar events easily 4) summarize email and attachment content 5) extract important links such as unsubscription, account activation, etc. 6) highlight good deals.
  2. Emphasize design principles, especially: simplicity, tolerance, ease, comfort, feedback.
  3. Novel and fit in the mental model. Balance the two within the 3 prototypes.

The three ideas I selected are a combination of many of the ideas of the above notes. They are the ideas that fit the above selection criteria. The details are described in prototype sessions.

  1. Down to earth, incremental change to the existing interface: smart email manager.
  2. VR to organize emails into themed rooms. A virtual assistant can help the user to do everything if the user wants.
  3. App reading emails to the user. Everything is through a voice interface.

Prototype 1: down to earth, incremental change

Figure 2: prototype 1

Figure 2: prototype 1

The first idea I prototyped is the down to earth, incremental change to the existing interface: use different tabs for different folders. Presort emails based on user behavior and the email content into folders. Allow users to move them among folders. Able to add email content to the calendar and make notes in the app while reading.

Figure 2 is the paper prototype of this idea. The paper drawings show the major functionalities.

The left panel demonstrates the functions of “folders”. The current tab is the “all” type, which pools all emails in different folders together. This way, it is very easy for the users to check whether the sorting is good, and sort the unsorted emails (NA) to the right folder. The circles at the beginning of the email line symbol which folder it currently belongs to. Each email has two parts in the tab view. One part is the “summary” of the title and content of the email. The other part is the useful links from the content if applicable. When it is a good deal, there is an exclamation mark to emphasize a good deal. In addition, the “+” sign at the tab line allow users to add a customized tab if so desire.

The middle panel shows that the user can move the emails to different folders if he/she thinks the current email sorting is wrong. Users can also add the email’s event to the calendar. These functions are revealed by sweeping the left and right of the email in the tab view.

The right panel shows the functionality of making notes while reading the email. This note session can be revealed by sweeping up from the bottom of the screen. It can be brought up in the tab view or the email detail view.

It satisfies all 6 functionality requirements mentioned in the selection criteria. In addition, it is simple, easy to use, and gives users the flexibility to tailor the interface to their own needs. This change is rather incremental changes instead of novel. The cost and effort to build is moderate. This prototype suits very well for the user profile (Figure 3).

Figure 3: Target user profile

Prototype 2: VR solution

The second idea I want to prototype is using VR to organize emails into themed rooms. I would like to use a textual prototype to represent my design.

  1. When the user enters the virtual reality, he/she is in a hallway. There are different paths leading to different locations. The locations are user-defined folders, only they are named with location related concepts. For example, “office” can be the place where the user’s work-related emails were collected. The “shopping center” can be where the deals are. In the hallway, the user can view all the emails and make changes to the sorting.
  2. Most of the interaction with the VR will be either voice command or gesture to move items around.
  3. Users can decide to have a “virtual assistance” which can take the shape of a butler or whatever the user chooses. Virtual assistance can help the user execute all actions possible in this app by listening to the user’s voice command. If the user decides to have a virtual assistant, the assistant will greet the user in the hallway. The assistant can also answer all user’s questions related to the interface or the emails.
  4. The user can see piles of letters on the office desk. He/she can unfold them. They can either read them or the letter can read itself to them (like in Harry Potter). The user can decide to review the full email or the summary.
  5. The links extracted from the emails can lead the user to the according to place if the destination website has a VR version. Otherwise, it will open a visual or voice interface to help the user finish the task.
  6. The user can reply to the email by simply voice recording and then can decide whether to translate them to text or directly send the voice recording back.
  7. The users can add events to the calendar and make notes like in real life. They of course can also do all of this through voice command or the virtual assistant.
  8. The user can change the sorting of the emails by folding them into paper airplanes and throw them to the correct destination, voice commands them to go to certain places, or ask the virtual assistant to do it.
  9. The users can explore the good deals by going to the “shopping center” or simply ask the virtual assistant to elaborate a good deal and make purchases.
  10. In essence, this is a virtual world where email contents are related to physical locations. The virtual assistant can also speak about the information and take voice commands to help users do everything.
  11. This interface also has the potential to help organize a user’s digital life and physical life that not limited to email.

This design satisfies all 6 functionality requirements. In addition, it is simple, easy to use and give the user the flexibility to tailor the interface to their own need. This is a revolutionary novel interface that will change the way people live. At the same time, the technical difficulty is very high. Current technology can only support basic functions. To build the best user experience, we still need some technological advancement, especially in the area of AI. This interface is going to be expensive to build.

Look back to the user profile (Figure 3), this solution would suit some of the users. The interface is expensive to build and the equipment is expensive to purchase currently. Thus the users need to have a high income. The technology is also cutting edge, so the adopters are likely to be tech-savvy population. The place to use this interface is also limited to relative private places such as morning in the bed or lunchtime. This is hard to use during meetings, driving, or chatting with others.

The solution is powerful in many ways, but also has limitations in many other ways to some users under certain contexts.

Prototype 3: voice interface

The third prototype I want to develop is the voice interface that can read all emails to the users. The user can interact with the interface through voice command. The major functions are tailored to the 6 requirements, which is very similar to the prototype 1.

Here I want to use the wizard of oz prototyping technique. Due to the limitation of the length of the report and time constraint, I am not going to write a whole script for every functionality. I will just write about a short scenario.

Context: morning, user exercising, checking emails.

User: Mailboxes, how many unread emails I have received since yesterday?

AI: From yesterday at 8 am, you have 80 unread emails. 3 of them are important, 32 of them are deals, 45 of them are social activities. Do you want to listen to them?

User: Yes. Please read the important emails to me?

AI: Sure. Do you want all the details, the headline summary, or newsletter style?

User: Headline summary.

AI: The first letter: Serena’s schools are hosting a donation party on October 13th, 2018. Please RSVP.

User: RSVP. I am going. Put on my calendar.

AI: There is conflict on your calendar. A night out with your husband.

User: Cancel that. Email him. We can go to the donation party together and have our date there. Next email.

AI: The second letter: from your manager. She didn’t understand the calculation in your latest report. She wants to schedule a meeting to go over it.

User: Put a meeting on our calendar. 30 minutes. Whenever we both have time today or tomorrow. Get the earliest one.

AI: Meeting scheduled. Today 3 to 3:30 pm.

User: Next email.

AI: The last important email: your water bill is due in 3 days.

User: Pay the bill in full.

AI: Voice recognition passed. The water bill paid.

User: Put a note in the note I share with my husband. The water bill paid.

This design satisfies all 6 functionality requirements although I haven’t addressed all of them in the above script due to the report length and time constraint. It is simple, easy to use, and provides the user flexibility to tailor the interface. Voice command is a brand new interface of the app, although not as cutting edge as VR. The technology is demonstrated by Amazon Alexa and Google Home, although the level of intelligence I designed is quite advanced. Without the high intelligence, the user experience will be less than demonstrated in the above prototype. The cost and effort to build is significant.

Users are getting more and more used to voice interfaces. I think this solution is suitable for most of the users in the initial intended profile. The cost to build is significant, but the user base is large. So the cost increase transferred to the individual users can be minimized. This interface is particularly suitable to contexts such as driving and exercising. It is less suitable to be used during meetings or around other people.

Overall, this is a powerful solution with unique advantages.

--

--