As large enterprises continue to focus on using big data to drive business they are hitting roadblocks. Data from customers and operations live in scattered locations; different databases in multiple clouds and on premise servers. These systems don't talk, prevent companies from taking actionable insights and from engaging with customers in a holistic way.

We needed a data platform where:

Data Engineers can provision, shape and manage cloud and on premise data. Where they can easily provision data sources for self service use across an organization, as well as manage the tasks of moving, shaping and transforming data sets.
Data Scientists can easily get access to the data needed for analysis. Where they can use that data in a variety of tools such as notebooks and model building. The ability to painlessly share the results of their work
Developers can create data and analytics services, building them into pipelines that deliver data to their organization via API, Kafka streaming topics or into operational databases.
Business Analysts can get access to analytics and data sets. Where they can be empowered to analyze data on their own, without requiring the help of a data scientist in their day to day work.

Mapping It Out

First I created a basic concept map organizing the capbilities and tasks required for the data portal.

I created different versions of high level site maps as a way to engage with development and offering management. The goal was to reflect a user's mental model of where data should live and what tasks they need to complete. We needed something that would make sense for the differing personas we needed to serve.

Design Concept: Project first view

Users of all backgrounds typically think of their data services and assets in terms of what project it's for. Such as, "I used this cloudant database instance and an ETL (extract transform load) data movement for Project X, and a dashDB database instance for Y." Users likely only work on one or two projects at a time, so in this concept I put project level filtering at the top level so it would be easy to switch views. Upon sign in, users would see the latest services and activities they had recently worked with. Switching to a project view would filter those services and activities on the same page. Our development team was interested in the concept but it was rejected because of concerns of all the API calls from each service that would be required to implement it.

Design Concept: Focus on Power User

This concept was aimed at the "power user" or productive use user; a user who regularly engages with the platform on a daily or weekly basis, is educated on how it works and, is looking to save time and do their job more efficiently. On login, they would see their most recent projects and system alerts; a user could quickly go to whatever project they are working on or actively respond to a critical system alert such as a failed data movement or a service which went down.

I also looked at how they could quickly move between content sections in a menu. A user is most likely working on this platform on a desktop, so I designed the menu for both on hover and on click states. On hover allows for an experienced user to quickly switch to a different section, when they recognize the icon and know exactly where they want to go. The fall back on click state, which combines both the icon and text link, is for a newer user who isn't sure where they want to go or when they haven't yet discovered the on hover state.

Because the platform was made up of a collections of different services, I proposed putting breadcrumbs at the top of the page, front and center. It would integrate with the site branding and make it easy for the power user to navigate the site without requiring use of the main site navigation.

Looking Forward

The data portal has gone through many iterations, and its current version will be launching in Summer 2016. The work I have shared was the beginning concepts; the product in its current form has changed. Upon formal launch, I will show the medium and high fidelity wireframes my team and I developed to get it to a finished product.