Empowering Sales Decisions through big data was the result of a semester long open-ended research and design project sponsored by BNY Mellon. The goal was to deep dive into one line of business in BNY Mellon - Global Markets, and through researching their specific needs, create a design that improved their work. Our need-finding included primary and secondary research, interviews, concept mapping, personas and collaborative maketools. Our process book can be found here The working prototype can be found here

My Role

Team Lead and Tech Lead

Team Members

  • Soleil Phan
  • Cristina Shin
  • So-Hee Woo


For the research portion of our project, we needed to learn more about our department, Global Markets. We decided to interview professors at Carnegie Mellon in order to gain more insight about this line of business.

We had an interview with Professor Bryan Routledge in Tepper to discuss Global Markets and financial markets as a whole. From this conversation, we were able to lay a foundation to our knowledge in this field.

Literature Review

Although we received invaluable information from these interviews, we decided to gain more insight through online research. We found that the role of the Global Markets department is to provide clients with services that offer growth and risk management. Because this line of business is spans on a global scale, we sought out to create a product that would increase the efficiency of this department.

Research tool / Maketool

We also created a collaborative maketool that we distributed to people within BNY Mellon. The purpose of the research tool is to gain more insight on the different roles within the Global Markets department. More specifically, the results will help us define the relationships between one person and another as well as the types of collaboration that occur around meetings. This led to a major pivot-point which led us to move our product in a different direction. Before we wanted to address information display and create a visualization tool for sales to client meetings, afterwards we decided to contextualize information relevant to the client-sales relationship and create a visualization tool for sales research.

  • maketool-07
  • maketool


After canvassing our research, we decided to focus on one persona, “the sales representative”. This persona was an archetype of the type of person who was on the client-facing side tasked with maintaining and developing business relationships. We used this persona to craft a credible scenario in which our product would be used.

Concept Map

Our team created a concept map to help us gain a full picture of the flow of information among all the key members involved in the business relationship. This technique help us craft the type of key interactions that would go into our final application.

  • conceptmap

Mock Ups / Wireframes

We created wireframes and mockups that were used to test the design among some users. We iterated on these designs until we felt comfortable in moving forwards with a hi-fidelity prototype.

Preliminary Design & Feedback

We interviewed two employees of the Global Market department to gather feedback from someone who would use our product. We asked them questions that would help us further in our design process. These interviews helped us add more granularity to the content that users would need to support their client.

  • info
  • discover
  • context
  • optimize

Working Prototype Development

I was responsible in creating the high fidelity prototype that we would present to our client. We decided to use HTML5 technology to allow for cross platform usage. Since banking still uses desktop computers for most of their work, it did not make sense to go with a native platform such as IOS or Android.

The development language we chose was Javascript, specifically NodeJS and ExpressJS as the framework for this prototype. NodeJS is an asynchronous non-blocking event-driven architecture that allows for scalable apps to be easily created. Since our focus was on creating a solution predicated on real-time updates and communication, it made sense to build on this platform.

We also used MongoDB as our database solution. In choosing MongoDB over SQL, the document-orientation structure and JSON language response made it a clear winner and suitable for a NodeJS application. MongoDB also has the ability to MapReduce data, which provided key functionality in our prototype.

As an open-source project, our source code can be found at