Design problem: Streamline dataset upload process
According to user testing during interviews, the process of uploading datasets was not clear ot the user. The original experience was vague, and lacked guidance and affordance
Improved experience must be able to support:
Upload 1 or more files (current max is 4)
Each file needs a description
Potentially able to autogenerate descriptions that a user can approve or edit
Each file needs a description
Potentially able to autogenerate descriptions that a user can approve or edit
Scenarios for the best solution
a. Is it better if the user uploads all files at once and adds descriptions after?
b. Is it better if the user uploads each file one by one, adding a description with the file?
c. AI Engine generates first pass at description that a user can accept or update it
b. Is it better if the user uploads each file one by one, adding a description with the file?
c. AI Engine generates first pass at description that a user can accept or update it
Design hypothesis put to test
I decided to move forward with the option that streamlines the user's experience by batching like-actions together to reduce cognitive load and context switching. Thus, the user would first upload all datasets together; and on the following page, the user would add descriptions to each dataset.
After speaking with the engineers, we decided to table leveraging AI to create a first pass of a dataset's description because the user has the most context and knowledge of their own dataset, especially at the beginning of the experience. AI can only know as much information as it is provided to it. We believed it would be more beneficial to do down the road, once the AI Engine better knows specific user datasets and recognizes patterns and common themes.
Final Design
Iterations Process and Thought Process
Design Components created
