Are visual analysis tools poised to become pervasive?

I spent most of last week at InfoVis 2008 in Columbus, Ohio. You might remember that I delivered the capstone presentation last year at InfoVis 2007, which also served as the keynote presentation for VAST 2007 (Visual Analytics Science and Technology). Last week the 2008 edition of this presentation was delivered by Christian Chabot, cofounder and CEO of Tableau Software. Chabot and I share the belief that visual analysis software is needed by a broad audience of people, not just those who have the term “analyst” in their titles. We also share the belief that with well-designed visual analysis tools like Tableau, visual analytics is poised to explode.

Participants in the conference consisted primarily of academics—professors and graduate students who spend their days inventing and refining visualization tools and techniques for making better sense of data. Chabot clearly wanted to challenge this audience to direct more of their efforts toward the practical needs of a broad audience of potential users.

Chabot identified four conditions that have set the stage for the current readiness of visual analytics to take off:

• Data explosion
• Technological advances
• General awareness
• Industry consolidation

The overwhelming amount of information that people now face has created a desperate need for tools that will help them make sense of it. Modern computer hardware and the Web have provided the infrastructure that is needed for people to interact with and share information effectively. Awareness of the visualization’s potential has reached a critical mass. Traditional business intelligence vendors, along with their tired, low-yield approaches to data analysis, have been bought up by large software corporations where they will languish, which has opened the door for better approaches to capture the attention of market. By rejecting the sins of traditional business intelligence vendors, refusing to compete for the hearts and wallets of customers through a litany of useless and ineffective pseudo-analytical features, software companies such as Tableau that are thoughtful, agile, design-oriented, and well-informed, have differentiated themselves from the pack and are now reaping the rewards of their commitment to give people analysis tools that really work. One result that we’re beginning to see is the gradual spread of data analysis tools to organizations of all sizes (from Google to the local bakery), and their proliferation throughout all parts of those organizations.

When the founders of Tableau Software were initially crafting their vision, they identified five core principles of visual analytics’ adoption:

• People adopt visual analytics primarily to help them see and understand complex data.
• People adopt visual analytics primarily to help them see and understand massive data.
• People adopt visual analytics primarily to help them see and understand new visual paradigms.
• People adopt visual analytics primarily to help them see and understand hidden insights.
• People adopt visual analytics primarily to help analysts save time.

Chabot is a Stanford MBA who worked for years after graduation as a high-end analyst—one of those guys that spend their days tackling complex analytical problems using complex analytical techniques. The other founders of Tableau, Chris Stolte, who earned his doctorate in computer science at Stanford by developing the prototype for Tableau’s eventual product, and Pat Hanrahan, the Stanford professor who supervised Stolte’s work, were immersed in the world of academic information visualization research. Their assumptions about what it would take to get people to adopt visual analytics made perfect sense, given their perspective at the time. As time passed, however, they kept their eyes open and learned that each of their assumptions turned out to be flawed.

Flawed Principle #1: People adopt visual analytics primarily to help them see and understand complex data.

Although sometimes complex, the data sets that people analyze are usually fairly simple. Chabot advised those of us in the information visualization community to start simple. Rather than focusing most of our attention on solving the complex, highly-specialized needs of a few, we can solve much more widespread problems that are just as important by making it easier for people to do the simple stuff that they must do over and over again each day, which are now unnecessarily onerous and time-consuming.

Flawed Principle #2: People adopt visual analytics primarily to help them see and understand massive data.

Although sometimes massive, the data sets that most people analyze are not particularly large. Chabot recommended that we start small, making it easy for people to work not just with huge corporate databases, but also with small files stored in Access and Excel.

Flawed Principle #3: People adopt visual analytics primarily to help them see and understand new visual paradigms.

Although there are times when new visual paradigms must be invented to solve peoples’ needs, most problems can be solved with proven visualizations, such as bar charts, line graphs, and scatterplots. Chabot suggested that we start proven by making it easier for people to use what we already know to work well in a seamless fashion.

Flawed Principle #4: People adopt visual analytics primarily to help them see and understand hidden insights.

While it is true that one of the great benefits of visual analytics is the discovery of previously hidden insights—those “Aha!” moments that we all crave—the primary reason, by far, that people want good visual analytics tools is more mundane, though no less useful: to save time. Chabot pointed out that we can design great tools that get out of the way, allowing people to become engaged in the act of thinking about data, rather than distracted by the mechanics of using the software.

Flawed Principle #5: People adopt visual analytics primarily to help analysts save time.

While analysts desperately need better tools to help them do their jobs, even greater benefit can be gained by providing tools that anyone can use, enabling everyone who must make sense of information to do their jobs and, as a consequence, freeing up analysts to spend their time solving the more complicated problems. With religious zeal, Chabot warned that we can no longer serve the needs of small groups with specialized needs, but should invite everyone to the table.

At the end of his presentation, Chabot reviewed his message and challenged us with these final facts:

• Millions of people need visual analytics technologies to help them understand information.
• The current state-of-the-art in business analytics (what most people rely on to do their jobs) is tragic.
• The primary barriers to visual analytics’ adoption are (1) awareness, (2) misperception, (3) ease of trial, (4) ease of deployment, (5) ease of use, and (6) ease of price.

What business intelligence vendors have still failed to do, a new breed of software company with roots in information visualization research, is poised to finally deliver. The world needs what we have to offer. To get it into the hands of those who need it, we must bridge the chasm that divides academic research and commercial software. Tableau and a few other ventures have done that. They’re inviting others to join them—not tomorrow, but now—because the time is ripe.

Take care,

Read More