“Good design” and “big data” are two phrases not often associated with each other. But yesterday, at the Strata + Hadoop World 2015 conference in San Jose, Calif., New Relic director of design and UX Etan Lightstone aimed to bridge that gap with a presentation titled Designing a Data-Rich UI, delivered in front of approximately a hundred highly engaged developers and designers.
Etan originally planned to call his talk “Designing for Data” and focus on data visualization design, but he realized that the concept of data visualization didn’t go far enough. You can’t create truly actionable data visualizations without considering the overall User Experience (UX)—so he concentrated on how to build a design process for data rich user interfaces.
“There is something incredibly compelling about taking enormous amounts of humanly incomprehensible data and presenting it in a beautiful picture,” Etan said, “But when it comes to data visualizations as user interfaces, these types of approaches aren’t very actionable.”
A quick exercise:
For example, take the interactive heat map showing measles outbreaks over the last century that was recently published in the Wall Street Journal. It may look pretty, but is it really the most useful visualization?
Maybe not. If the goal is for epidemiologists to be able to track the spread of the disease, that’s a geographical issue, Etan noted. This visualization puts the states in alphabetical order by state, making a geographical analysis very difficult. Perhaps a time-slider controlled map might have been a more suitable approach.
All about the process
Let’s take a look at the processes involved in designing a data-rich UI. This workflow shows a typical data visualization design process:
But creating a simple and intuitive UI requires more than this, Etan explained. “You need to learn exactly who your users are, and what problems they are trying to solve. Our goal as designers should not be to provide a picture of the data, but to provide end-users with ‘understanding’ so that they can make better decisions.”
That means UX designers need to combine this data-focused process with a UX design process where we research our customers and their goals, synthesize conceptual UI through a scenario-based approach (narrated use cases), and finally craft mockups and prototype just enough so that you can get early feedback and iterate on it.
Using research to develop personas—archetype users—can help us make choices and inspire us to design and build the right stuff, Etan said. It’s fundamentally about understanding your end-users for your product… their goals, challenges, behavioral traits, and tasks. “As you interview potential or existing customers, you basically look for patterns that inform how they might use your product. And you can start categorizing them if different clusters emerge.”
Hint: You can’t always directly ask users what they’ll want… they’ll often just list out all of the data they need.
The next step is to craft scenarios—uncovering the user’s context and flow to help make UI decisions. It’s increasingly common to see UX design borrowing from Hollywood, presenting the interface as a narrative storyboard, focusing on users and their context as much as on the interface itself.
Hint: Begin by depicting your end users in their existing systems without your imagined solution, highlighting their points of confusion.
In general, Etan said, it’s often a mistake to think too much about your data set and specific UI elements before theorizing the overall UI that would make use of it. That can lead to “designer tunnel vision,” moving you too quickly toward a presumed UI solution that may not be right for the actual data and users. The goal is to create the ideal customer workflow—that may be impossible, he said, but the process will hopefully inspire good choices.
Making it real
“Prototyping is enormously hard to do right,” Etan said. “How do we build just enough that we can go learn what we need to learn through customer feedback, and still have the freedom to throw away the prototype without wasting too much time and money?”
That feeds into the next step, evaluation and user testing, which is ideally where you iterate and loop back the most, Etan said, because “making huge changes to partially built production software later on can be much more expensive than something intended to be a throwaway prototype.”
Here at New Relic, we try to follow these processes to create products that feature beautiful and actionable designs and user interfaces. With dedicated designers on every product team, we listen and work with our customers to create designs that continually delight them.