IBM Watson Data Platform kickoff to launch (now called watsonx)
Working with data is silo-ed. Data professionals work with multiple, non-collaborative tools and publish their findings in multiple locations. In a single large enterprise, you could have two business analysts in different business units needing the same data to analyze. Not only is it hard to find, but once they have access, they have to pull it into multiple to tools for analysis that may or may not work together. The potential to share data or work on the data with integrated analysis tools in a collaborative setting is lost.
My role & team
I led the design team that built 1 of the 3 tools within the platform, Data Catalog, and led the work for the main integration experiences that unify the tools within the platform– Projects, Community, Data Services, and creating a unified UX pattern library. Those integration experiences helped users collaborate, work across multiple tools, and connect or access third-party data and storage.
​
The design team included: Bhavika Shah (Design Manager), Kacie Eberhart (UX/Visual), Noelle Hoffman (Visual), Tom Workman (Front End Developer), Amanda Hughes (Visual), Frances DiMare (Research) working with 4 PMs and 4 development teams across Canada, US West Coast, India and Japan.
The goal
The goal of the platform was to enable teams to collaborate with confidence, share access to data — no matter where it lives — find new and unexpected insights, and deploy machine-learning models that deliver business-changing results. It was built to solve the following painpoints with working on big data:
​Pain point 1: There is not one tool/product that provides a seamless end to end experience for data professionals.
​
Pain point 2: Data provenance, data gathering, and data cleaning are still the key areas where data professionals feel the most frustration and spend the most time, from finding data to work with in general, figuring out where the data came from and whether or not it is valid, and transforming and cleaning data to be in a workable state.
​
Pain point 3: Data professionals publish to varying locations or only share to their immediate chain of command, there is general repository for sharing findings.
Data Professionals
We designed Watson Data Platform as a solution for all users that work with data from storage to governance to analysis.
Data Engineer: Data engineers are problem solvers. Their role is to design, build, and maintain pipelines so that the company’s data can be found, accessed, and used.
Data Scientist: Data scientists are investigators. Their role is to leverage big data to create and apply algorithms that surface actionable insights for the company to use.
Business Analyst: Business analysts are cultivators. Though technical abilities vary, they use the data, tools, and teams available to them to produce the best actionable business insights for their company.
Chief Data Office(r): The Chief data office and officer is an emerging role that ensuring an organization’s data is governed in such a way that it is compliant with industry regulation and accessible to teams across the company to enable analytics.
Connecting services and building integration components
In order to build Watson Data Platform, multiple design and development teams came together to create tools that worked with each other. In December 2016, the portfolio had about 20+ products that did not work together and had with their own UIs. While the development team worked on the APIs to get the tools to talk to each other, design worked on the other major challenge: How to unify 20+ UIs?
In order to overcome this challenge, design teams began working on 3 tools that tied together a majority of the functionality within the platform– Data Science Experience (DSX), Data Refinery, and Data Catalog. The platform then tied these tools together using integration components so the experience and look for all three applications were the same. We did this for 2 reasons:
1. Don’t want to design or build things twice
2. Don’t want to have two versions/experiences for the same action
While the Watson Data Platform overall design team consists of over 25+ designers. My team led the work for the Data Catalog tool and the integration components working with the other design leads and design teams to coordinate efforts.
In order to connect the applications, we designed 4 major experiences that unify the tools– Projects, Catalog, Community, and Data Services– along with creating a unified UX pattern library that all teams contribute to. The 4 major experiences are how users collaborate, work across multiple tools, and connect or access third-party data and storage.
A deep dive into designing the Data Catalog
You might be asking, this seems like a ton of work, what did the process look like?
In December 2016, a team of developers had created a user interface for a product called: Data Hub- a data catalog but since the team did not have a design team everything built as features without a coherent user flow. Our job was to asses the current state of Data Hub to identify how to transform into a true Data Catalog. We highlighted 4 areas of work: develop a consistent navigation, simplify pages that had repetitive actions and content, reduct the 9 ways to load data, and clean up user permission hierarchy.
​
The design team hosted multiple workshops with the development and offering management teams to discuss the current issues, the technical restraints, and the future direction of the product. The entire team went through a persona exercise, creating user flows, and having the development team and product management create conceptual wires. At the end of the workshop, we created a plan to redesign and rebrand the experience and migrate users over to the new platform.
​
We created heuristics of the current experience, built out multiple user flows, tested with existing users, and iterated on the fly to get to the final product.
Launching Watson Data Platform
Watson Data Platform began in December 2016. By October of 2017 it went beta, and in December 2017 it launched to general availability. This aggressive delivery timeline meant that design, development, and product had to work in close communication. Designing on the fly, designing with development, designing using only assumption were all challenges we faced. By restructuring the way the broader Watson Data Platform team communicates and works together helped overcome these challenges.
​
While we had limited time to conduct user research, we tested the platform UI with 8 enterprise companies as sponsored users in order to get insight into how the platform can be used as a whole and drive direction for the future iterations. Since then, it has transformed into watsonx.
IBM Sponsor User
“...there is nothing comparable out there which use single unified platform for the personalized roles within an organization... business analysts, data engineers, and your data scientists. All these three people who are going to deal with your data."
IBM Sponsor User
"It's an end-to-end data management platform starting from getting your data 'til the end. The whole journey is taken care of in Watson Data Platform."