In-Depth
Research Excerpt: Turning Business Intelligence into an Enterprise Resource (Part 2 of 2)
What organizations must do to transform BI from a departmental resource to an enterprise one.
- By Wayne Eckerson, Cindi Howson
- 08/24/2005
Once they understand the benefits of enterprise business intelligence, many executives want to deploy a solution immediately. Unfortunately, it often takes several years for an organization to gain the experience and skills and build the technical infrastructure required to deliver an enterprise solution and reap its benefits. TDWI’s Maturity Model defines six stages that organizations pass through on their way to a mature BI deployment. Only when organizations reach stage five—the “adult” stage—can they begin to roll out BI on an enterprise scale.
Although a discussion of all the steps required to achieve BI Maturity is outside the scope of this report, we will highlight a number of tasks that organizations must undertake to transform BI from a departmental resource to an enterprise one. Much of the advice below involves aligning the BI solution with business strategy and making sure the business, not the IT department, owns the solution and guides the IT project teams during implementation and beyond.
[Editor’s note: Last week we presented the first half of this task list. This week we include the remaining six items.]
8. Leverage Dashboards and Scorecards
For a majority of users—those who view and interact with reports on a regular basis but not intensively—Web-based dashboards are the “new face of BI,” a perfect medium for delivering information in a way that matches most users’ requirements.
Dashboards offer an intuitive, multi-layered interface that delivers customized views of information and structures navigation into detailed data so users don’t get lost or waste time searching for the right report. The top level displays exception conditions using graphical indicators that alert users to issues they need to explore. The second layer provides analytical functionality that lets users slice and dice data to explore issues in a guided fashion. The final layer consists of detailed data and reports. In essence, dashboards provide built-in, guided analysis customized to individual users or their roles.
“We have 500 detailed reports that contain 85 percent of all the information people might want. However, we discovered that many users couldn’t find the right report and got frustrated. We now use dashboards that provide multiple layers of summarized information on top of reports. This creates a pyramid in which users access information at the top and gradually work their way down into the detail,” says Michael Masciandaro, director of business intelligence at Rohm and Haas, a specialty materials and chemicals manufacturer.
9. Develop Functionality and Reports Rapidly
Users often abandon business intelligence tools if they don’t see their requirements for new functionality or reports delivered in a reasonable time frame. To address this problem, many organizations create networks of “super users” to create reports for users in their workgroups or departments and give them new codeless development tools, which TDWI calls analytic development environments (ADE), (see note 5) to accelerate the creation of custom reports and applications. ADEs make it easy for developers or power users to create custom reports or solutions in a matter of hours or days, depending on the complexity of user requirements and the availability and condition of the data.
The super users not only create reports for their colleagues, but also provide training and first-line support. In turn, the IT department trains and supports the super users on the new tools and serves as a secondary help desk. The IT department creates business views or metadata layers that allow these users to create reports without having to understand SQL or the physical database schema. These formalized networks of super users ensure that BI tools stay in the hands of business users and keep IT out of the business of creating custom or ad hoc reports.
Turning over ad hoc report creation to departments also seeds the enterprise with innovative techniques that the IT department can disseminate to the rest of the organization. For instance, the technology group at Dow Chemical is monitoring departmental efforts to build custom dashboards to see what techniques are transferable to the enterprise. “The trick is to capture the creativity and flexibility at the departmental level and then scale up the solution for use elsewhere,” says Gregg Reitz, supply chain BI product manager for the company. Dow Chemical was an early adopter of enterprise BI and standardization, having successfully standardized on its BI platform in the early 1990s and implemented its solutions globally.
10. Deliver Fast Response Time
Most users are impatient. They won’t wait long to obtain reports or get query results. Ideally, they want to query information at the speed of thought; in other words, instantaneous response time. BI administrators need to tune databases and leverage caches to ensure reasonable response times. They also need to educate users about the nature of the data they are querying. For example, users who request a custom report containing large amounts of detailed data can expect to wait much longer than they would for navigating hierarchies in a standard report.
“We set up processes to keep performance within bounds,” says Bikram Kalra, manager of decision support at NBC Universal. “We only let power users create ad hoc reports, and we develop standard reports that everyone else uses. These reports contain prompts, so one report now replaces dozens of older custom reports. And we cache the top-level reports to provide fast response times, but drill-downs are dynamic.”
Michael Masciandaro at Rohm and Haas, which has standardized on a Web-based BI tool for the majority of its 3,500 BI users, says, “It’s hard to get users to use the tool if they don’t get instant gratification; they lose patience and leave. Our goal is to have every click [get a response in] five seconds or less. A high percentage of responses are faster, usually less than a second, because we cache data sets and reports.”
11. Deliver Right-Time Data
You can get the majority of users to endorse a BI solution if it delivers timely information that enables them to make better, faster, and smarter decisions and fix problems before they escalate into red ink on the corporate income statement. Consequently, most organizations are now moving to update their data warehouses daily, if not more frequently, to deliver actionable information to users. Right-time data delivers actionable information to users when they need it, usually on an intraday basis. (See note 6)
For example, Rohm and Haas moved from monthly updates of its data warehouse to daily and sometimes hourly updates after it standardized on packaged software from a single vendor for both operational and analytic applications. Besides providing business users with actionable information, this has enabled the company to shrink the time it takes to close its month-end financial books from 10 days to two days. It accomplishes the rapid updates by using packaged content, more efficient ETL processes, and changed data capture techniques.
12. Monitor Usage
The best way to judge the effectiveness of a BI environment and training programs is to monitor usage. Some companies view usage statistics as an early warning system for problems. For example, International Truck and Engine Corporation tracks BI usage during the test phase of a new BI report or set of functionality. “If only three people out of 10 are using the system, then we meet with the other seven to find out the problems they have with it and make changes before we roll out the release,” says Jim Rappé, group leader of enterprise data warehousing at the company.
Rappé’s group tracks usage statistics so closely that it now knows what the uptake rate should be after issuing a new release of the software. If adoption rates are lower than normal, the team jumps into action. “If usage is below the norm, we book a 30-minute presentation during a departmental meeting to provide additional education and answer questions. We try to be proactive,” says Rappé.
13. Continuous Iteration
It’s one thing to manage a BI project that is delivered on time and within budget and meets user requirements. It’s an entirely different thing to sustain an initial success over the long haul. To do this, it’s critical to maintain close relationships with end users, continuously gather requirements, and deliver new iterations on a regular basis. Many BI managers who have successfully steered a project to completion change roles and become in effect BI evangelists, who sell and market the value of the resource and continually gather new requirements for future releases. (See note 7) Some organizations hire account managers whose primary job is to serve as liaison with business users in a division or unit to identify how the BI resource can be leveraged or enhanced to meet existing or emerging business goals and strategies.
Summary
There are doubtless many other best practices when it comes to delivering an enterprise BI solution. The key is to remember that BI is not just a set of query and reporting tools; it’s an entire information management and delivery environment. Although considerable technical work is required to deliver an enterprise BI solution, the business must:
- Provide leadership for the initiative
- Take responsibility for standardizing data elements and changing processes that improve data quality
- Devise a road map that aligns BI efforts with overall strategic objectives and delivers the most value to the broadest range of users within the organization
NOTES:
5: Wayne Eckerson, Development Techniques for Creating Analytic Applications, TDWI Report Series, May 2005. http://www.tdwi.org/research/reportseries
6: Colin White, Building The Real-Time Enterprise, TDWI Report Series, November 2003. www.tdwi.org/research/reportseries
7: Wayne Eckerson, “The BI Evangelist,” TDWI’s Case Studies and Solutions, June 2004. www.tdwi.org/publications/newsletters