The proliferation of relatively easy-to-use data visualization software means the tools are getting to the hands of more business users. But using these tools is not like adding instant lemon aid mix to water. Organizations considering visual analytics and data visualization projects should keep a number of things in mind, according to users and analysts. (For more on data visualization technology trends and vendors, see “Advances in Data Visualization Software Empower Business Users.”)
The following are seven pieces of advice when considering data visualization packages:
1. Make sure business users can get insights when they need them.
Boris Evelson, a vice president and principal analyst at Forrester Research, says businesses have to be willing to recognize the need for good-enough data that’s timely in the majority of cases. Take customer segmentation data as an example. It may take a day or a week too long to extract data from “a single version of the truth” when business users need insights right away.
Opening data to more widespread exploration and analysis using visualization programs forces this issue. And, rather than rely on IT, “business has to own business intelligence. Success depends on agility,” he says.
2. While the tools come with many visual display options, set design standards for your company.
When it comes to design, organizations should establish standard visual metaphors. “The wow factor is nice, but you need consistency,” says Shawn Spott, vice president and manager of marketing research at the Royal Bank of Canada’s (RBC) U.S. Wealth Management subsidiary.
Evelson says that most business users need to have the graphical design of their reports and presentations guided by the applications. “The business person shouldn’t have to know which types of data are best shown with pie charts, a heat map, a scatter plot” or other graphical formats.
3. Evaluate whether to use specialized visualization tools or those included with enterprise analytics and business intelligence packages.
A key purchase consideration is whether to go with a separate data visualization tool, when the latest versions of enterprise business intelligence suites have visualization components that aren’t far off the standard of the specialized vendors, and indeed, the big enterprise players often add products through acquisition. It may come down to flexibility, performance, ease of use or the type of data being accessed.
4. Know which business metrics you want to measure before embarking on visualization projects.
Visual analysis and data visualization applications should also support the calculation and measurement of business metrics. Potential customers should have a clear sense of what metrics they will need to focus on, according to Evelson.
“For example, customer profitability by individual customer or by region, time, product line, or sales territory, and how often do these parameters change? These factors determine whether the implementation should be SQL–based, or in memory–based, with free exploration, with or without any limitation,” Evelson says.
5. Consider whether you need a data visualization package to connect to existing reports.
Users also have to consider the software’s ability to find an existing report, Evelson notes: “It’s one thing to build a new report on a nice clean model data source, but if you’re connecting to a new, unscrubbed data source, how do you do it on your own, without the help of IT?”
6. Plan to incorporate data visualizations into business processes.
Despite the appeal of visualizations as a cool project to work on, the displays are not an end in themselves, says Remco Chang, assistant professor of computer science at Tufts University.
“For example, it should be used to identify patterns, and then maybe you can write a data mining rule to catch a certain occurrence” Chang says. “Once you model it, the problem shifts into the known.”
7. Don’t forget the human factor.
Crucially, because visual analytics and data visualization are decision support tools, organizations need to consider the human factor. “You shouldn’t focus on visual interfaces and presentations, which are well understood, but rather on human biases and how people interact with computers—the integration of human analysts and machine learning and automation,” Chang says.
Interface design and visualization options can also distract from the results of the data exploration, analysis and presentation, Chang says: “It’s a matter of how that extends people’s ability to extract meaning.”
Ted Smalley Bowen is a freelance writer and editor based in the Boston area Reach him via email at firstname.lastname@example.org.