Build a product for Data Scientists to augment their process and allow them to exponentially test, run, and analyze more data and models, resulting in more actionable information for them and their clients.
Led design for a startup product within BCG, that was developed for data scientists (specifically for data scientist consultants at BCG) to exponentially streamline and systematize their processes. Oversaw the UX and UI design and implementation of new features and functionality that resulted in increased usage and more productive cases for users. Working closely with the head of product we implemented consistent and organized research in order to surface roadmap features, analyze them, create, and plan quarterly epics for the product.
The implemented features resulted in thousands more runs per day, support of government and other high profile programs, and ultimately Source.AI was acquired by DataRobot.
Industries Big Data, Consulting, Machine Learning, Data Science, Agile Data Science
Roles & Services Design Lead, User Experience (UX), User Interface (UI), User Research, Information Design, Functional Design, Roadmapping
It was an exciting challenge and experience to come into the project with little knowledge about Data Science, ML, and AI. The immediate first steps were to get grounded in the current product, understand why decisions had been made that led to the features and functionalities that existed, and what the current identified challenges/problems were. In addition, learning about Data Science, interviewing and meeting with users, team members, developers, engineers, and the principal product manager, all built the foundational understanding needed to push forward, to start hypothesizing, and to build the roadmap.
Sketching, Wireframing, & Prototyping
As with all projects, initial work starts with a piece of paper and pencil where ideas take shape, structures created, wireframes and flows plotted, and user needs defined. Throughout the process of iterating from sketches, to wireframes, to mockups, and to finished products, stakeholders are part of the conversation the entire way. So when hand-off of the finalized assets, prototype, and/or annotations to the dev team, everyone is already on the same page, and overseeing implementation is much smoother.
Prototyped Scheduling Functionality
When initially brought on board, running a model was a manual process, only being able to run one at a time. After communicating with users (champions within the organization), scheduling runs was prioritized. Wireframes and ultimately the high-fidelity designs were created in Figma, utilizing a functional prototype to communicate the feature and functionality. This was then handed off to the dev team and used to communicate annotated functionality and designs.
Sitemap & Analysis
In addition to providing real time design support for current needs, we also conducted an initial audit of the current product, its information architecture (IA), the actions that a user could take, and how everything connected. This was the beginning stages of reorganizing the IA to better fit how data scientists work on cases and projects. This also helped surface gaps, inconsistencies, and opportunities across the product (which we color coded as well to help highlight).
User Research Processes & Roadmapping
At every step of the way, we were in communication with and conducting one-on-one interviews, surveys, and other data/feedback collection. In order to streamline user research, which was coming in from multiple sources, we consolidated data into a single source in order to surface features and functionality that could improve the product. Part of this endeavor we created personas for each case type and linked those to current cases, current users, and connected all the research to real-world users. Once surfaced in the airtable we could push the feature to the roadmap and design.
Navigation & Structure
From that research came the beginning stages of the redesign of the information architecture, navigation, and search and filtering functionality, to enhance user experience and usability of the product. There were roadblocks for the Data Scientists to utilize the product in an efficient manner. By leveraging user research and understanding how data scientists work, a more flexible hierarchy and structure for data scientists to execute more efficiently was being designed. This was a preliminary wireframe/mockup.
Better Products for Data Scientists
Redesigning features and functionality, and including new more clear and concise organization of core functionality was serving the purpose of simplifying the data scientist’s work while enhancing/augmenting their capabilities. We didn’t get to the testing rounds of the redesign within the Data Scientists’ workflows due to the acquisition unfortunately. Ultimately, this would have let them switch in and out easily, jumping around and implementing the workflow that was most efficient and productive for them. A customized experience with almost no customization needed.
Learning & Moving in the Right Direction
Part of what made this project so great to work on, aside from the incredible team, developers, product manager, and users, was the challenge of understanding and taking a complex and complicated work flow, and trying to make it more seamless for the data scientists. We were moving in the right direction, the roadmap had some excellent features down the pipeline that stemmed from extensive research, but we did not get to implement due to the acquisition (a bittersweet end result for all that were working on it).