This year's list of fatal business intelligence (BI) flaws, identified by Gartner Inc. last week, isn't that different from those of past years. The difference, however, is the intended audience.
"There's a whole new crop of recruits coming into BI," Bill Hostmann, an analyst with the Stamford, Conn.-based research firm, said in a phone interview from Orlando. "I'm here at the Gartner Symposium, where we have about 7,000 folks, and a lot of my contacts are with CIOs that have done BI in the past but have done it in very one-off ways," Hostmann said.
CEOs are increasingly expecting enterprise-wide -- versus departmental -- BI deployments to help improve strategic and operational decisions. That means CIOs, rather than line-of-business managers, are now in charge.
But with the benefits of enterprise-wide BI deployments come enterprise-wide BI deployment flaws.
"It's pretty typical as an organization moves through [BI] maturity, they're going to have failures," Hostmann said. "You're going to have learning mistakes. You may fall into them, but that doesn't mean you want to stop your BI initiative. That's not an option."
Get the business talking
One of the most common flaws when deploying BI enterprise-wide, Hostmann said, is IT embarking on the BI initiative without enough input from the business. The result is often a first-rate BI system that nobody uses because "its business value" hasn't been communicated to the rest of the organization.
Similarly, too many organizations fail to set out a comprehensive BI strategy – identifying how BI is expected to improve business operations and, ultimately, the bottom line -- before setting out with an implementation.
The solution for both mistakes, according to Hostmann, is for CIOs to create BI competency centers (BICC), sometimes called BI centers of excellence. A BICC should not only streamline BI and data management processes throughout the enterprise, it should also bring together both IT and line-of-business leaders to map out expectations and agree on a unified strategy and tactics to achieve them.
Business intelligence begins and ends with data
Poor data quality and a lack of common data definitions are two more potentially fatal flaws, according to Hostmann. The problems are only exacerbated when pulling data from siloed, departmental applications and systems from throughout an enterprise.
Unfortunately, too many organizations are loath to address both issues.
But once data quality and data definitions are in line, the work's not over. Hostmann warns CIOs to beware of executives and other workers who "dance with the numbers." He said too many people "mess around with [BI] calculations" when the results are not what they expected or contradict long-held beliefs.
"There has to be a culture of accountability," he said. "You have to agree on the metrics, you agree on the information you're going to use [to arrive at] those metrics, and then you don't touch the calculations or the presentation of those metrics. They are what they are."
To help do so, key performance indicators, business rules and analytics calculations should be embedded at the system level, Hostmann said. That way, managers have less opportunity to manipulate the data.
Keep your options open
When choosing a BI system for an enterprise-wide deployment, Hostmann said, it is important to evaluate vendors other than your highest-profile business application supplier. CIOs shouldn't take the path of least resistance and simply go with Business Objects' BI platform, for example, because you also run SAP ERP and HR applications.
Evaluate third-party vendors, both large and small, to find the best fit for your organization. Integration and compatibility among vendors is fairly high, Hostmann said, so using one vendor's BI software and another vendor's database is not as daunting a challenge as it once was.
"The world of mix-and-match is very much a part of BI platforms," he said. In fact, your main IT vendor's BI capabilities "may not fit with your requirements."
He also urged companies to build BI competency from within. Understanding internal data is what ultimately gives organizations a competitive advantage, and that's not something that should be outsourced. "You've got to own the competency, the program management, the skills development," Hostmann said. "That has to be a core internal competency."
Functions like extract, transform and load and data warehousing management can be outsourced because they aren't competitive differentiators. "What differentiates you as a company is how you use that information to make decisions and manage your performance," Hostmann said.
Finally, CIOs must prepare themselves and the rest of the organization for change. Rigidity and BI is simply not a winning combination. "BI applications are going to evolve because the business is going to change," he said. "The kind of information and analysis that people do is going to change. So you have to have people's expectations set that this is going to take a while."