Data Refinery – Big data preparation



Company- IBM

Product- Data Refinery (part of IBM Watson Data Platform)

Year- 2016, 2017

Role- UX Designer & Researcher

Skills Used- User Interviews, Cognitive Walkthrough, Competitive Analysis, Heuristic Evaluation, Journey Maps, Concept Diagrams, Sketches, Wireframes, Working Prototype, Usability Tests




IBM Data Refinery is a self-service data preparation and integration service for users to access and prepare their data for analytics. It is part of IBM Watson Data Platform which aims at tackling the challenges of big data. It enables data-driven professionals to work together in a simpler way to quickly find new insights through cognitive-assisted machine learning. The most vital part of the software is the “Refiner” (shown above) where the data is cleaned and prepared.

Some of the projects I worked on include data metrics, data transformation operations, new user experience, usability tests, etc.




UNDERSTAND- Meet with the project manager and all the stakeholders across different teams to understand the requirements, development and time constraints and the future phases for that project. I would also prepare a UX evaluation of the current state of the product. If multiple sessions are involved, I would draw concept diagrams and flowcharts to facilitate the discussion and clarify my understanding.


RESEARCH- Ideally, I would like to do user interviews but due to budget and time constraints, this usually meant talking to internal data scientists, engineers and analysts to understand their experience and clarify any technical questions. Occasionally, we would do generative research to refresh our personas and learn more about their work process.


DESIGN-  I would start my design process by maybe making a user journey map based on interviews and my evaluation. I ll then brainstorm, come up with ideas, sketch them out, do competitive analysis and look at how other products might be solving similar issues.

Once a direction for design is finalized, I would start making mid fidelity wireframes to design the details and go through many design iterations based on feedback from stakeholders and guerilla testing. I would make sure that my designs are using the existing patterns if any and are as modular as possible to reduce development time. Once designs are approved, I would explain everything to the visual designer and hand it off to him/her to do the high fidelity designs. Sometimes I might create some of the high fidelity designs as well if the visual language for the components in my design is already defined.

Data Metrics Designs  (Password: IBMDR2017!)


USABILITY TESTING- Depending on the project and budget, I would design a usability test, prepare discussion guides, recruit users and do moderated remote usability tests.  I would then synthesize, find insights, present them, prioritize the issues discovered based on severity and the effort required to fix them, and even design solutions for some of the problems discovered.

Here is some feedback I received on Data Metrics which is one of the functionalities I designed:

“Impressed with the summarization”

“Very Helpful. Definitely something I would use.”

“Much more robust than what I was thinking.”