Operational BI Application Deployment PracticesOperational BI Application Deployment Practices

VentanaView™

Research on the deployment of business intelligence into business operations indicates the need for specific practices for delivering operational information. Various factors now drive enterprises away from static report-based information delivery to ad-hoc reporting and interactive query-based applications. When deployed to large user communities, these operational BI applications can have significant implications for hardware infrastructure and user support. Ways to shorten turn-around time for user information requests top IT priority lists. Support issues that could be addressed ad-hoc for small departmental deployments require more role, process and technology structure. Performance assurance shortcuts which limit data access provide tactical relief, but may stunt overall application ROI. Ventana Research recommends the use of ROI and TCO assessments as part of the planning process to determine critical success factors.

Assessment
An ever-increasing onslaught of data, coupled with ever-increasing competition, is forcing decision-makers to seek technology that enables management by exception. Dashboards and scorecards are leading implementation forms through which BI is used to support management by exception. Enterprises also are deploying BI technology not just for management decision-making, but also decision-making by front-line workers involved directly in business operations. In 2004, Ventana Research conducted a survey of organizations that are deploying operational BI applications. This article discusses usage modes, needs and challenges uncovered by that study.

The research showed that, while most organizations still embrace static reports (53% of respondents), parameterized reports (48%), dashboards (32%) and other interactive BI use modes were also used frequently. Through other discussions with BI managers, Ventana Research observes a clear trend for more future projects with dashboards and the desire to reduce costs for supporting static reports by moving to parameterized report interfaces. Movement from static, batch-run reports to parameterized reports and ad-hoc queries may reduce batch window size and maintenance support, but may also require significantly higher performing hardware. User support may also increase as interfaces get more complex to support parameters and ad-hoc query development. The hidden trade-off for reduced IT support is the need for hardware and software and an increased learning curve burden for users. IT organizations with casual or part time support processes for BI will need to more carefully plan and provide for these kinds of deployment changes.

The research also showed that for existing systems, the foremost desired improvements on organizations’ operational BI systems are twofold: to ease development/modification of reports, dashboards, and scorecards, and to provide more information to users. These two needs go hand in hand, as more information to end-users drives the need for more reports, dashboards and/or scorecards. Simply put, the business need for information is ever-increasing, as is competition for information. Organizations with inefficient processes for designing, developing, testing and deploying BI are actively pursuing ways to streamline those processes while lowering costs. More now than ever, careful cost (TCO) and efficiency analyses will be necessary to assure successful BI deployments.

Organizations pursue a variety of approaches to assuring query performance within their operational BI systems. Some are intuitively obvious: faster hardware, caching of data, various tuning techniques, etc. However, research also showed that IT organizations also frequently constrain access to data as a tradeoff against meeting performance criteria. Limitations include constrained report parameters, limiting access to smaller user groups, applying query governors, etc. A tactical perspective may result in complacency regarding access constraints as an acceptable performance assurance policy. Nevertheless, line-of-business managers need to understand that limiting access may reduce overall operational BI system ROI, especially if effort has been made to gather data that isn’t optimally accessible. To justify providing broader access, organizations need to quantify the value of data access to assure they are gaining maximum value from their BI assets. While difficult to do, assessing data value is a critical step to driving BI strategy.

View
Large-scale operational BI applications increase in complexity along multiple dimensions simultaneously. User scale, data scale and calculation scale requirements all must be met, but can work against each other, resulting in significant system design and deployment challenges. Ventana Research recommends that organizations that intend to deploy large-scale operational BI applications first assess the adequacy of their application design, deployment and support processes and assets. These organizations should clearly understand the purpose of these applications and how they will provide an investment return. ROI, TCO and other similar investment evaluation methodologies should be employed pro-forma to provide project objectives by which to measure project performance.

Eric Rogge is VP and research director for Business Intelligence & Performance Management at Ventana Research.

Ventana Research is the preeminent research and advisory services firm helping our clients maximize stakeholder value with Performance Management throughout their organizations. Putting research in a business and IT context we provide insight and education on the best practices, methodologies and technologies that enable our clients to leverage assets to understand, optimize, and align strategies and processes to meet their goals and objectives.