Data-driven decisions: The View from the dashboard
Early versions of dashboards were called executive information systems and had a similar goal, but they were not connected to original source data. They were derived from various databases and required significant input from the IT department. In fact, they were not sustainable.
Today's dashboards draw directly from data warehouses or even from multiple databases, and are far more interactive. "Organizations have become much better at creating a smooth data surface for dashboard reporting," Mollot says. "The ERPs and data warehouses are much better as source transaction systems." The technology to allow users to drill down and ask a series of related questions is robust, and customers are gaining competitive advantage from it.
Another difference in more recent dashboards is that they can be set up to present different views depending on the user's role. "Dashboards are often designed to cascade," says Mark LaRow, VP of products at MicroStrategy. At the top level, data is rolled up to show the big picture, but depending on roles, functions and levels, other users see a smaller slice of data. Once the system is set up, the underlying data can be used to provide a concise snapshot to a wide range of recipients with varying needs.
At Corporate Express, a supplier of office and computer products, MicroStrategy's BI solution allows 2,000 employees to monitor business performance at various levels. Known for its ability to deal with large databases, MicroStrategy's product operates against a 4.5-terabyte Oracle data warehouse. The executive team relies on a dashboard to monitor customer buying trends and enterprise performance. At Lowes, MicroStrategy provides merchandising and executive dashboards.
Usability is a key factor in the success of dashboards, according to LaRow. "You should not underestimate the value of convenience and clarity." The right information needs to be in the dashboard, naturally, but the dashboard should also engage the user.
"The real power of dashboards," he adds, "is that you can take multiple sets of data and array them on a screen that leads you to want to see another set of data right next to it." For example, an HR director looking at a spike in attrition might want to check out demographics, salary, length of tenure or other factor in attempting to explain the trend. The dashboard should allow this interactivity.
Fine-tuning the interface should be deferred for a while, though. "Don't try to make the interface pixel-perfect right away," advises LaRow. "The dashboard is likely to evolve over the first four to six months, then stabilize. Spend some time on the iterations, then finalize the look." Developers can get bogged down in making the dashboard attractive, deflecting too much attention from its functionality.
Like any decision support tool, dashboards must be used wisely. Some important determinants of success that operate within a firm cannot easily be quantified, and therefore are rarely measured. Failing to incorporate those factors into the decision-making process can send an organization down the wrong path. Other forces operating outside the company, while quantifiable (such as interest rates), may not be reflected in the dashboard, even though they may exert an influence greater than any under the organization's control. But to the extent that what can be measured can be managed, dashboards offer a valuable option for helping companies cope with the data avalanche.