overview
Imagine having to manually scroll through 1000s of pages to find answers to a few small questions everyday. The goal of my team's project was to make this experience more enjoyable and intuitive. Since majority of the research and designs are being used today by IBM to develop the tool further, I am not able to share most of the artifacts of this project publicly on this website. However, for those who are more interested I am happy to set up a call to walk through the design thinking process further. Below, you can find a simple version of the design process we followed.
my role
Project Lead:
-
Conducted 3 rounds of user interviews
-
Created interactive prototype with 30+ screens in Sketch and InVision
-
Completed user testing and lead iteration process.
context
6 week long project for IBM Patterns design internship program
Team size: 5 designers + Sponsor team
tools & skills
Sketch, InVision, Mural
the design process
01. the problem
initial problem statement
How might we establish a framework for IBM to popularize future research by creating an engaging and enjoyable experience showcasing an AI-based natural language interface to users?
initial blockers
-
Given a tool, and had to find a problem this tool could solve
-
Understanding the AI tool, it’s capabilities and limitations
-
Deep dive into researching IBM’s AI tools
-
-
No initial sponsor users to interview
-
Finding sponsor users
-
the existing technology
-
Difficult to understand what the technology was or how it worked
-
Unclear use cases
-
Too much info was being presented at once
-
Not engaging
questions & assumptions board
Purpose: To understand problem space better and by breaking it down into what you know and what you don't know. We narrowed in on the highest value questions and determined high priority assumptions
success metrics
mindshare
must be an example of innovation and raise awareness of the work done at IBM
functional
Prototype must deliver to the user an engaging and enjoyable experience in the process of data extraction from documents
02. research
user interviews
Purpose: To understand the main type of user that would benefit from this tool and to build empathy, uncover deeper issues, and target concrete problems to solve.
1st round: data scientists (5 interviews)
​
-
What types of data sets do you usually analyze?
-
"PDFs are final artifacts. There's no analysis that comes from there."
-
​
-
Can you describe your current process when filtering and sorting through data.
-
​"85% of my workload is finding, then analyzing the data. A tool like this would reduce my workload by maybe 5%.”
-
​
Key insight: data scientists are not our main user.
2nd round: students & industry professionals (12 interviews)
​
-
Student pain points:​
-
Have to search for information in 300 page pdfs/books.
-
Are equipped with large datasets that are often just pictures/pdfs.
-
-
Industry Professional​​
-
similar pain points to students​
-
potential to be a revenue generating client
-
​​"We work with many institutes that have pdf-form contracts. The ability to analyze those contracts and extract information from tables could be pretty powerful for us."​​​
-
-
​
​​
​
Key insight: students and industry professionals would be a good target, novice user for this tool
our research and analysis led us to...
updated problem statement
the design challenge
How might we show off the AI tool in an engaging way to let potential users experience and become aware of the amazing technology coming from IBM? ​
03. ideation
empathy maps
Purpose: To gain a deeper insight into the main pain points of our average users and putting ourselves in their shoes by breaking down the pain points into what the user is doing, saying, thinking, and feeling.
industry professional persona
student persona
-
Extracts information from documents for job purposes
-
Needs robust data extraction tool to improve workflow
-
Can help demonstrate tool's functionality as enterprise solution
-
Extracts information from documents to complete assignments
-
Fits the "can be used by anybody" model
-
Easy to recruit for interviews and testing
storyboarding
persona: meet steve
thinking
"There's probably a better way to do it, but I don't know what it is. I usually would just hand it off to someone else.”​
about
-
Newly promoted Senior Audit Manager at Accounting Firm
-
Recent MBA graduate
-
Husband and Father
context
-
Reviews the accounts of companies and organizations to ensure the validity of their financial records
-
New to this role, and feels a lot of pressure to do well
-
Current workflow takes hours and works overtime
-
Finding information is not easy
feeling
doing
-
Manually going through large financial documents with millions of lines of data
-
Uses control-F where possible
​
​
-
overwhelmed
-
stressed
-
cannot balance life
-
helpless
sketches
04. framing the solution
needs statements
Purpose: Identify key needs of target users that will be addressed with the design solution. We broke down all research and ideation into these statements to make sure we were addressing both the stakeholder and the user.
industry professional
-
Needs a way to extract unstructured data from dense documents so that he can simplify his workflow.
-
Needs a way to ask questions in natural language so that he can save hours in his search for information.
-
Needs a way to experience the tool in a simple and interactive way.
stakeholders
-
Functional - Engaging and enjoyable experience in the process of data extraction.
-
Mindshare - must be an example of innovation and raise awareness of the work done at Watson Innovation
-
Collect user data - Gather (optional) information to collect user demographics and gauge interest.
hills
Purpose: To communicate the intent for the project with meaningful user outcomes. Hills frame problems as intended user outcomes, not predetermined implementations, empowering teams to discover breakthrough solutions.
1. A student/professional is able to quickly search through unstructured data using natural language without having to manually go through each page.
​
2. A student/professional is able to properly understand the output of their unstructured data search using AI driven in-context assistance.
​
3. A student/professional is able to learn about the tool and global table extraction, through an interactive web experience
05. iteration
user testing
Purpose: Identify problems in design to make sure the solution is fitting the needs of the user and to remove bias from designers that might have slipped in through the design process.
key insights
-
Majority of users skipped of description of technology on landing page --not interested
-
Want to add animations to make it more engaging for user
-
-
Directions in demo are repetitive and numbers in steps are distracting
-
Added progress bar at the top with one word directions instead of steps​
-
-
Return to start button placement is off and is very small
-
Moved the placement and made it more visible on the demo​
-
-
Many people liked simplicity and minimalist visual design
experience-based roadmap
Purpose: Explain the future of the product and what goals further iterations of the product would try to meet.
the cupcake
what you can enjoy now
the birthday cake
what you can enjoy a little later
the wedding cake
what you can enjoy in the future
06. reflection
This process helped me learn a lot about leadership in product design. As the design lead on my team, I led everyone in the group through each step of the design process and worked collectively to meet each deadline.
​
Through this experience, I learned the most important part of working in a team is communication and being able to have everyone felt heard. My team was able to get along very well and we each shared our perspectives very openly. When disagreements would arise, we learned how to effectively collaborate to come up with the best solution so everyone was satisfied with where our project was heading. I also learned that the main reason disputes arise is due to miscommunication and having organized meetings can significantly reduce the number of such disagreements.
​
I also learned the importance of referring back to user-centered design whenever disagreements arise because it is important that the product is effectively designed to meet the needs of the user specifically rather than the people on the design team. Empathizing with and understanding the users needs is one of, if not the most, important aspect of design thinking as in order to make truly impactful solutions it is imperative to make sure you are meeting the needs of the user specifically.