Responsive Web

Data Modeling Platform

2018 - 2019
UX Designer
Project Type
Launched Project for Responsive Web
Data Modeling Platform

Complex & Costly Data Modeling and Managing Process

Clients in the financial service industry are not obtaining enough value from their data. They lack a consistent, accurate view of the data across data providers, and also the resources, systems, and expertise to manage, normalize, clean, and aggregate the data. In response, Data Platform aims at realizing value from business data by offering tools to operate and maintain data. However, because of its complex and rigid data modeling process, and its lacking operation features to manage the service, lots of effort from the implementation and operation team is required, which may lead to an exponential increase in the cost of man-day when business scale up.

We spent 4 months researching users’ feedback for the current system, needs, and expectations, and 5 months to iterate and validate a business-oriented, user-centered, and automatic platform for data management.


User Research: Goals, Expectations, and Pain Points

We started by interviewing internal users to investigate how they were utilizing the product and their goals, needs, and expectations. After the first discussion, we developed further questions to understand the reasons behind the current system’s structure and design. By 3 rounds of interviews and focus group discussions, I generalized the jobs-to-be-done and personas.

The Implementation Team

The implementation team works for onboarding new clients. Their work in the old system was mainly concerned with 4 sections: Element, Table, Mapping, and Modeling. The element part was like a dictionary, storing the basic definition of data columns. The table part utilized elements to define data tables. The mapping section connected tables together. And in the modeling part, users added complex transforming rules to the mapped tables.

The Operation Team

The operation team works for running and operating data models for multiple clients. Their work involves 3 sections: Check Rules, Event, and Report.

Research Findings


With the new design, there’s no need to separate two groups as the implementation group and operation group. Internal users can switch easily between these two roles in different project phases. When onboarding a new client, more internal users can switch to be implementors, and at other times they can all be operators. External users, the clients, can now easily create and run workflows by themselves, and thus gain more value from this platform.

Besides implementors and operators, the system would also need internal user groups to maintain data, client, and user groups.


Concept Development

After investigating the existing product, we found that most work is about building models to generate reports and run models as required by clients. So we decided to introduce the “Rule-Task-Workflow” structure to simplify the data modeling process. With workflow building function, users only need to select tasks at each step and define conditions if necessary. Also, we iterated on the idea of operation management with task&workflow driven scheduling. Users should easily monitor running workflows with real-time alerts and messages.

Positive Tone and Active Voice

To manage the text we used in this system, I constructed a library of “section-page-scenario-headline-content-button” for the glossary and the text we used in all messages. I adopted a positive tone and active voice to make the message more clear and straightforward.


After validating concepts and information architecture with clients and the development team, I created and developed prototypes. I went through 3 rounds of low fidelity, and 4 rounds of high-fidelity. My low fidelity prototypes focused on concept validation and simple usability. My high-fidelity prototypes are for content display, visual design, and usability testing.

Usability Testing

We conducted 2 rounds of user testing to test key functions and data table features respectively. In the 1st round of user testing, I designed 3 tasks and 12 interview questions to test key functions of workflow display and workflow manipulating with 3 user groups - users from the implementation team, operation team, and external senior manager group. Though the interaction we designed is novel for the users, most of them completed each task in 3 minutes.

In the 2nd round of user testing, I developed 4 sections to know how users interact with a data table, which is the most commonly used component in our product. The test focused on how they search for items and how they commit actions to items. The result shows that users generally prefer column operations like in Excel.

Final Design

Based on the user testing result, I iterated on the UI design mainly for dashboard, workspace, entity management page, event management page, and the error display.