Data visualistion has languished behind its perception as a scientist's toy or a nice tool without a market. Computing power, demand and availability are starting to surge, pushing the tool to leap to the forefront but there are still some barriers to cross. Steve Ulfelder reportsIf you're a client of Deltek Systems and your account is in arrears, be aware that when the company managers and directors meet, your name shows up on a computer screen in big red letters.
This sounds primitive, but like many businesses, Deltek, a professional services automation company in the US, is just scratching the surface when it comes to data visualisation.
The technology, rooted in scientific applications, is now being merged with statistical analysis software. The goal is to replace reports and tables with powerful, eye-catching images that convey important statistical data to even casual users.
Users of the technology rave about its ability to help business people grasp the pertinent points in huge quantities of data quickly, and experts say data visualisation will quickly be merged into standard data analysis tools. But users complain that visualisation products still have some way to go where ease of use is concerned.
Until recently, most corporate information technology managers viewed data visualisation technology as a toy for scientists or as a nice business tool that couldn't find mainstream traction. But experts say that's finally changing, for three reasons.
First, computer power has finally caught up with the technology. "A few years ago, you needed a $US20,000 Silicon Graphics workstation to use visualisation," says Don Campbell, Cognos vice president of information delivery products. But that isn't the case anymore; new tools can run on desktops in a corporate environment, he says.
Second, the demand for business data is fearsome - and it's growing all the time. Even the most hard-core bean counters, born in Lotus 1-2-3 and raised on Excel, must work hard to pull the significant or potentially threatening numbers from spreadsheets. Data visualisation makes those numbers impossible to miss and easy to grasp by everyone.
"These types of tools can help you more quickly adjust your mind and pinpoint information without having to interpret it," says Bob Moran, an analyst at Aberdeen Group. "It helps you see relationships by looking at a chart."
This is a vital point. Experts say data visualisation software's ability to accentuate the relationships among data points is one of its main benefits. Moreover, e-commerce has put a premium on real-time data. In some sectors, it's critical to keep an eye on your own site traffic and your competitors' in order to see who's winning, what promotions are working, where the traffic's coming from and so on.
Data visualisation makes this possible. Rather than having to wait for reports or compare sterile columns of numbers, it's now possible to use a browser interface to gaze in real time at your vital e-business numbers.
The final and perhaps most significant reason for data visualisation's growth spurt is that vendors with deep roots in data analysis software - including Cognos, Visual Insights and SPSS - are building the technology into their product lines.
Keith Gile, an analyst at Giga Information Group, says data visualisation will follow the data mining trend. Just as data mining has faded into the background as a stand-alone discipline but is more widely used than ever as the backbone of customer relationship management applications, data visualisation will be folded in as the expected interface for statistical-analysis software, he said.
From analysis to synthesis
Deltek uses Cognos' Visualizer 1.5 to synthesise analyses gleaned from other Cognos tools into an easy-to-understand presentation. "It lets us learn which projects are most profitable and which regions have the highest concentration of clients," says Shimi Minhas, Deltek's director of business intelligence. "You can immediately ID those hot spots, then drill down to the detail."
Deltek has used Visualizer for about six months, Minhas says. Because the company is a long-time user of Cognos' data analysis tools, Deltek didn't consider competing vendors' products. Regional and project managers liked stepping up from spreadsheets and reports because "they say they can make faster decisions", Minhas says.
Paid-up accounts appear in green, 30-day accounts in yellow and delinquent in red. "In looking at a Visualizer [presentation] of accounts receivable balances, we can quickly ID which clients are not up to date," Minhas says.
Although Deltek is happy with Visualizer, Minhas says, it can be difficult to understand for end users unaccustomed to other Cognos products. In addition to such Cognos data analysis products as PowerPlay and relational database management systems supported by the vendor's Impromptu, Visualizer supports flat files and Excel files as data sources.
Visual Insights seems to be emulating Cognos' model. Founded in Bell Labs as a research project, Visual Insights started as a tools company but is fleshing out its product line, according to Michael Tatelman, vice president of marketing and business development. "We're now an analytic applications company using visualisation as a core differentiator," he says.
Moran says he agrees that Visual Insights is "moving into analytic tools very quickly".
A US energy company Enron, found that risk management was a natural application for data visualisation. "Risk management involves complex mathematical principles, and the results are not intuitive," says Rudi Zipter, Enron's director of market risk management. Enron develops energy and bandwidth portfolios, which it then trades like commodities. "We like to analyse and decompose risk across the portfolios," Zipter says.
Zipter's department started using Visual Insights' Advizor/2000 late last year. He says it "results in a keener insight of risk" because it "allows you to see where risk is coming from". Recently, the risk management department was certain that the risk in a certain portfolio was coming from one source.
But, he says, "by putting traders' portfolios in a 3D visualisation, we could see right away" that the problem area had been misidentified.
Advizor/2000 accepts standard Excel tables as input. It also accepts online analytical processing 'cubes' - structures that store multidimensional information - that are created in Microsoft's SQL Server 7.
Zipter does have a few suggestions for improvements to Advizor/2000. He says the flexibility built into the product brings an almost overwhelming number of possibilities when Enron develops applications. "And there's not a lot of documentation," he adds, "so it's learn as you go."
A more granular approach
Another data analysis heavyweight, SPSS, offers nVizn, a Java-based developer's tool kit that lets businesses create their own visualisation applications.
The advantage of the development-tool approach is its flexibility. SPSS's Dan Rope, nVizn's chief architect, mentions its "granularity" when discussing the absolute control developers can hold over their presentations.
"The developer then uses his knowledge of the [specific business] domain" to make data do exactly what he wants it to do, Rope says.
Waratah Corp, which provides software development and data mining services to the health care industry, chose nVizn when Health Hero Network, asked it to create an information visualisation component for Health Hero's Web-based patient care management system.
According to Michael O'Connell, president of US-based Waratah, patients at home answer a series of questions about their health on a daily basis. That data is then used to build a 3D visual representation that makes it easy for health care professionals to monitor patients' health on an ongoing basis.
The SPSS visualisation is superior to simple charts, O'Connell says, because "you're able to see trends in information across time with meta data - that's much more powerful". Danger signals such as weight fluctuations raise a red flag, triggering an intervention, he explains.
The trade-off for such granular control may be added complexity. SPSS officials say it typically requires a team of three to create a custom visualisation application with nVizn: a "domain expert" (in the health care example, this might be a doctor), a developer and an analytics expert.
So while Cognos and Visual Insights are attempting to build front-to-back data analysis suites that feature a visualisation interface, SPSS is selling a do-it-yourself kit that requires a more significant development effort up front but can potentially pay off with super-customised presentations that better fit an organisation's data and business.
While the business case for visualisation tools is powerful, vendors have their work cut out for them. Their push today is to overcome visualisation's reputation as a high-maintenance tool for scientists and to get software into the hands of subject-matter experts - the people who can best use them.
Tatelman says he believes the trickle-down of visualisation software will mimic that of desktop publishing tools.
Judging from interviews with users, however, vendors must first make data visualisation software more intuitive and easier to use. Right now, users appear overwhelmed by visualisation's unlimited possibilities. But this is a stage with new technologies and they always get through it, and what was initially intimidating quickly becomes well understood.
"That's a natural part of [any new technology's] evolution," Gile says. "Business intelligence software has always been focused at the power user. Everybody's got to see a real neat visualisation app before it catches on."
It's critical that products become easier to use and to populate with data. One significant problem for IT is that data visualisation tools have always been parasitic by nature - that is, they've taken data from existing programs such as spreadsheets and prettied it up.
End users saw the results and often demanded visualisation tools of their own. But where was the data going to come from? That was IT's problem. Now the melding of analysis tools and visualisation appears to be addressing that issue as well.
$US795 per client; site licence availablewww.cognos.comAdvizor/2000Visual Insights$US 295 per userwww.visualinsights.comnViznSPSS Inc$US10,000 per developer seat plus deployment licence; price varies with deployment sizewww.spss.com