Creating a Business Intelligence Culture

Slide Show

11 Guiding Principles for a Successful Business Intelligence Implementation

For some time now there's been a lot of buzz about companies getting business intelligence into the hands of more users, with the aim of getting the right information to the right people at the right time (or words to that effect). The conventional wisdom: Companies are limiting the potential usefulness of BI by making it available only to specialists, who create reports from centralized data and make those reports available only to select decision makers.


Not everyone agrees with this, of course. In 2009 I interviewed Nigel Pendse, a principal with OLAP Solutions who had just authored that year's version of the BI Survey, a report produced by the Business Application Research Center. He told me vendors were pushing the idea of so-called pervasive BI simply to sell more software licenses. While many folks require access to operational information, few of them actually need to perform analysis on it, he said:

A BI tool is for somebody who makes decisions with some latitude. If your decision-making is completely constrained, you don't need a BI tool. I just went on a trip. I certainly wouldn't want pilots getting their information about fuel from a BI tool! The plane has operational systems to show them what's happening. I don't want them sitting there doing analyses about optimizing fuel usage. Somebody at the head office may do that, but the guys flying the plane don't need to worry about it. Or even at the airport. There might be thousands of people working there, but I can only think of maybe a hundred who could conceivably get any value out of a BI tool.

Yet several companies featured in a recent Computerworld article are seeing tangible results from getting more users involved with BI. At 1-800-Flowers.com, users with access to real-time sales data created a quicker checkout process for fast-selling items, which the article says likely reduced costs and increased customer satisfaction. Bobby Nix, director of BI and analytics at consumer services company Allconnect, credits a 26 percent sales increase in 2011's first quarter to newfound staff expertise with BI, saying it helped sales associates focus on the best sales opportunities. And the Cincinnati Zoo & Botanical Garden saw a 30.7 percent per-capita increase in food and beverage sales from October 2010 through the first quarter of 2011 that the director of park operations attributes to BI.


The key to making this work, say the sources interviewed in the article, is creating a culture that values BI. Like so many bits of good advice, this sounds simple but can be maddeningly tough to execute. Many pieces on BI make the case that business users, rather than IT, should lead the initiative. On the face of it, this makes sense. At many organizations, IT still assumes sole responsibility for BI. Not surprisingly, this often results in solutions not well suited to business needs.


The problem with this approach, of course, is that it positions IT as being subservient to business users. Thanks to its high-level understanding of an enterprise and all its systems, IT can and should play a key role in helping foster a holistic view of data and how it can be used in ways that benefit not just individual business units but an entire organization. IT must convey this message in a way that stresses business benefit. As IT Business Edge's Loraine Lawson recently suggested:

If IT wants to stop enabling data silos, it's time to stop talking about data and master data and "product data" and so on. Instead, talk about information (isn't that what the "I" in IT stands for anyway?). Talk about "intelligence." (Who doesn't want to be James Bond and collect "intelligence"?) Use real words that are more holistic than the oh-so-techie terms IT tends to use.

Encouraging business users to think in broad, big-picture strokes won't be enough. IT must walk the walk by adopting a similarly holistic approach to data integration, she wrote:

... Once, data integration was enough - but now, it's not. Business users don't just need you to integrate the data from one set with another set. They need information integration - they need to know, for instance, that you've not just combined databases, but you've resolved the semantic differences so "apple" means "apple" and not "pear." ... Data management is already on the right road with the trend toward "enterprise information management" and an information-oriented architecture. Semantics and meta-data management will be key. Embrace those trends.

In the Computerworld article, IDC analyst Dan Vesset advises using the appropriate tools to make different types of decisions. Rules-based, automated systems offer the most efficient way to make tactical decisions. Corporate decisions usually require conferring with colleagues, which is facilitated by using tools with collaborative elements.


Some other good tips from the article:

  • Create a BI/analytics practice within IT and corresponding liaison groups in each business unit. At 1-800-Flowers.com, these groups "complement each other perfectly," says the company's CIO. The article doesn't specify, but I assume members of these LOB teams have other responsibilities as well. Some companies may want to appoint a full-time team of tech-savvy business analyststo work with data architects, technical architects and other IT staff to ensure the technology framework can support business requirements, an approach I wrote about earlier this month. They are responsible for gathering requirements, developing BI roadmaps, managing BI budgets and overseeing BI and data governance programs.
  • Determine existing BI abilities. Ask each business group to identify who needs access to BI and to classify these folks as basic, intermediate or super users. The goal at 1-800-Flowers.com is to "turn basic users into intermediate users and intermediate users into super users," the CIO tells Computerworld. IT assumes responsibility for providing training. I shared some other suggestions on BI training from publisher Meredith Corp., which recruited super users from among the business ranks, trained them on the more advanced features of the company's BI tools, and encouraged them to share their new knowledge with coworkers.
  • Let users do it themselves. At Allconnect, IT staff "spend a lot of time training and mentoring" business people on analysis techniques, but they don't actually create analyses unless they require complex analytics, according to the article.
  • Educate everyone. Recruit influential business evangelists for BI, but make sure the message gets out to everyone in the organization. At the Cincinnati Zoo, even part-time cashiers understand the importance of gathering patrons' ZIP codes for data analysis. This approach ensures a practically constant stream of ideas on how to use data to improve the business, the zoo's director of park operations tells Computerworld.