Even in IT departments that have a resident business intelligence analyst, that person is most often building production reports upon request, says Few. "But typically they don't understand the data. They don't really know how people are using the data they're putting in that report," he says.
"There's a disconnect between the people who actually work with data to make decisions and the [IT] people who supply the data they need," Few continues. "Finding people that really understand the data and understand the technologies that organization has to distribute the data -- finding those two things in a single person -- is relatively rare." (For other necessary skills, see Qualities of a good data visualizer.)
What Few and others are talking about is the more finely honed aesthetic sense that today's data visualization requires. Boris Evelson, vice president and principal analyst at Forrester Research, says there are two levels of data visualization skills emerging. One level refers to a person's ability to use the latest technology and tools to analyze and present information. Rather than using Excel or even Cognos, for example, data analysts are using Tableau or Spotfire to create more visually pleasing and more easily comprehended charts and scatter plots. (See
22 free tools for data visualization and analysis
for more suggestions.)
But that's not enough in some applications. Recently a large New York City bank told Evelson it needed someone with deeper skills to visually present a sophisticated and comprehensive portfolio analysis -- analyzing thousands of clients with various types of investments and risks. Although the bank had "all the right tools and technologists," he says, it was looking for someone with a specialized understanding of how the brain reacts to and digests visual information.
"It was not about the technology of data visualization, but the psychology of visual perception," he says. The bank wanted someone who would know which types of visualization techniques work best for different types of data, as well as the limitations of certain techniques. For example, "a significant proportion [about 7%] of the population is colorblind," he notes. "So maybe they shouldn't exclusively rely on color."
(Article continues on next page)
Dataviz: A brief how-to
Know your data, know your audience, and determine the message you want to communicate.
Reduce the data to what's needed to communicate your message, remembering that, without context, numbers mean little.
Determine the best means of expression.
- Some quantitative messages are best communicated with words, some with tables of numbers, some with specific graphs -- bar, line, scatter plot, etc. -- and some with a combination.
- These principles aren't intuitive; they require training into how our eyes and brains process visual information. Consult a data visualization expert on this step (or train yourself or your staff).
Design the display to communicate simply, clearly and accurately.
- Don't include anything that isn't data unless it's needed to support the data.
- Avoid unnecessary color variation and visual effects, or even grid lines in a graph when they aren't needed.
- Make non-data elements only visible enough to do their job; they should never overshadow the information.
- Visually highlight information that's most important to the message.
Suggest actions in response to the data. Most quantitative messages aren't presented merely to inform but also to motivate a useful response.
Source: Stephen Few