American Express
Cornerstone Dataprep

Cornerstone is American Express’ global big data ecosystem and a critical enabler of innovation and growth of the company. We are centralizing data across the company with strong built-in governance, enabling democratic access to a broad user community and optimizing company platforms through new synergies.


Cornerstone Dataprep is a web-based platform that promotes data adaption by enabling data citizens to get hands-on with data extracts and assets simply and efficiently.


4 Months

My Role

Research, User Flows, Information Architecture, Wireframing, UI, Prototyping.

High-level Goals ✨

My Process

Problem Discovery

Since day one at American Express, I've heard many complaints from reviewers, vendors, coworkers, and beta testers about data accessibility. I worked closely with my product manager, on pinpointing the key issues our existing users were having.

I lead journey mapping workshops with stakeholders and the other product designers to gain insight into our user’s experience.

My Product Manager and I also decided to shadow powerful data users taking part in different stages of the data submission process to find out where our users were having the most trouble.

User Research

The user research was great for quickly gathering information from a sample of my target audience, while the interviews allowed me to gain a deeper understanding of the participant's behaviors. The main takeaways from the two methods were:

The user research was great for quickly gathering information from a sample of my target audience, while the interviews allowed me to gain a deeper understanding of the participant's behaviors. The main takeaways from the two methods were:

  • Heavy dependence on the Technology teams or Advanced Analytics teams to provide data to the business.
  • Barriers to finding, understanding, and accessing data lead to a long lead time for decisions driven by data insights.
  • Current data accessibility is not sufficient to meet today's dynamic data needs.
  • Users need more actionable and intuitive data visualizations.
  • Competitive Analysis

    I conducted an in-depth competitive analysis of our product. This revealed an opportunity for differentiation from a product strategy perspective: most analytics platforms do not equip their users with the resources to work on available data within such big organizations. It also informed the way I would later design the information architecture, data visualizations, and user experience.

    Problem Statement

    How might we empower our users to self-service data access for business data citizens through a secure, well-managed environment where users can quickly find, understand, and provision data?


    By then I had already gathered enough information to get started with the wireframes. I created a set of mid-fidelity wireframes of all of the key screens needed to complete the main user tasks that I will be testing.

    With the information architecture solidified, I engaged in a process of making low, mid, and hi-fidelity wireframes with intermittent prototyping, testing, and iterating. Given the lean development team, it was important to design a system that would be easy to build in a short timeframe.
    My primary considerations were:

    Data visualizations: Show users only the information that was actionable and intuitive

    UX/UI:In designing the system, I ensured that it was comprehensive and could withstand large company growth because of its scalable structure

    Mid-Fidelity Prototypes

    After doing discovery work, I started on user flows and wireframes. I focused on the areas we wanted to change and quickly incorporated them into an InVision prototype so that we could get the concept in front of beta customers. Through these first rounds of testing, we found additional areas for improvement.

    User Testing

    UX Researcher and I set up six usability tests to understand how people would react to the platform for the first time.

    Usability Test Tasks

    • Preview Data (5min)
    • Enter Table Edit View
    • Open data details and filter missing values from column_name_2 (5min)
    • Remove column_name_5 (5min)
    • steps interaction plus Save & Exit (7min)


    • The Steps panel wasn't noticed until after a couple of tasks were performed.
    • A user wanted to right-click on the tile to enter Table Edit View.
    • Logic actions make sense, but it takes time to figure out the actions.


    Instead of the small popup screen with registration form fields, we moved to the new design system style using a fullscreen layout. We also decided to add descriptors, help tiles, and instant field checker to help our users correctly fill the form.

    We relabeled the action tiles to include a visual representation of each action.

    Visual Design

    The final result of the platform was a system that addressed the pain points I had discovered during research. I designed an interface that met technical limitations but prioritized usability. I also designed it to withstand large company growth because of its scalable structure. The redesigned platform received positive feedback internally and from our returning customers.


    We started by sketching up the new user registration process which utilized a single registration process to provide access to most of the data. This was to replace the existing process which required multiple registrations to access multiple data sets.Thanks to that the user could now register only once and then work on their data without fear of having to request access to another data set.

    Filter by Date


    Table Edit



    We had worked out the bugs we uncovered during the beta, implemented the new usability standards, deployed new features and flows, and set up Mixpanel analytics to track user behavior to dig into greater insight.

    • Increased engaged users by 12%
    • Increased data accessibility by 8%
    • Public launch of desktop app

    Working with developers

    I presented my research findings and redesign to the rest of the team and worked with the dev and product team to integrate the usability improvements over the next few sprints. I learned a lot from our lead dev about how the product was built and how our databases interacted with the site. We explored how to handle all possible use cases in onboarding and needed security and data policies for each case.

    My takeaway

    My biggest learning was how to utilize qualitative and quantitative research methods to validate a product idea. As the sole designer on this project, I was able to juggle user research, stakeholder interviews, and prototyping to create a product strategy that met the needs of our users.