Summary

In our recent benchmark research on Operational Business Intelligence, we found that operational workers require more diverse data sources than analysts who use traditional BI tools; these operational sources include desktop spreadsheets and flat files. In fact, when we compared these findings with those from earlier benchmark research (“Business Intelligence for Operational Performance,” done in 2005), we found that operational data stores (ODSs), data warehouses and data marts are no longer the dominant sources of data; spreadsheets and even flat files now are more common. Some organizations may conclude from these findings that they need to expand the range of data sources integrated into their enterprise data warehouse or ODSs to manage access more effectively. Others may choose to provide data access directly to those sources or use enterprise information integration software systems for federated access. But no matter what the information integration and access strategy, it is clear that for daily decisions, operational workers need to view relevant information wherever it resides.

While organizations have good reasons to extend BI to their operational employees, Ventana Research believes that the task is not as simple as replicating to new user populations what has worked for traditional BI. In that model, technically astute user communities analyze data from carefully chosen sources and require updates to the data less often than daily. For operational BI, however, the largest percentage of research participants indicated that daily or more frequent updates are essential. We believe that organizations need to pay careful attention to the development of their information technology and management infrastructure to ensure that it meets the requirements of operational BI users. Organizations should evaluate vendors’ abilities to integrate BI tools with their information infrastructure, especially in light of recent mergers and acquisitions that have reshaped the competitive landscape in both BI and data warehousing.

View

BI and data warehousing are firmly established as the methodology and technology platform of choice in many organizations to support reporting, data access and analysis. However, our operational BI benchmark research (sponsored by Business Objects, Cognos and InetSoft and media sponsors BusinessIntelligence.com, DM Review, Intelligent Enterprise and IT Business Edge) found that while plans to expand BI are nearly universal, there will large obstacles to overcome to make such plans a reality. One of the primary impediments continues to be the integration of data from disparate sources. Accessing, using and sharing data all first require that users be able to integrate it regardless of its source. In the research, 35 percent of our participants cited difficulty in integrating data from multiple sources as one of the strongest points of dissatisfaction with current operational BI deployments.

Operational workers in very large organizations (that have more than 10,000 employees) need access to more sources than those in smaller companies; 29 percent of participants from these organizations said that it is important to access to more than 20 sources. While overall the highest percentage of participants (34 percent) said that access to five to 10 sources is adequate, we found that future deployments for particular user groups will involve increased numbers of sources; 40 percent of participants told us that the number of sources their front-line workers need will increase from two to four currently to between five and 10 within a year.

As they expand BI to operations, organizations must address the confusing and difficult-to-maintain array of custom code typically used to handle integration within and between applications and data sources owned by different business functions. In our research, 70 percent of participants said that they prefer to centralize operational BI deployment rather than use departmental or federated alternatives. Centralization allows organizations to build operational BI on top of an information hub, such as an enterprise data warehouse, that integrates data from disparate sources. Organizations have been able to use the data warehousing approach to eliminate much of the point-to-point custom code used for traditional BI. Our benchmark research found that they intend to take this approach to resolve integration problems they encounter during operational BI expansion as well.

Along with standardizing data integration and transformation in a data warehouse, the centralization option often involves establishing an enterprise BI and reporting platform at a level above the warehouse. This technology can decrease costs by reducing redundancy in the organization’s operational reporting and data access capabilities. However, just lowering costs is insufficient reason for choosing a centralized approach; the driving force should be to deliver high-quality information to operational workers who can use it to gain important business benefits. It is critical for organizations to note that in diverse operations and business functions, not all workers have the same BI and reporting requirements. Centralized deployment should not simply become “one size fits all” BI and reporting. The platform should support flexibility in data visualization, the implementation of performance management methods and processes, and user collaboration through portals, e-mail and emerging social networking options.

Assessment

Our benchmark research found that the primary business benefit participants seek from operational BI deployments is access to existing data sources. Therefore, creating a strategy for reliable, scalable and high-quality data and information integration should be a focus of any BI expansion to the lines of business and operational functions. Centralized deployment will help organizations address data management and integration issues in a cost-effective and standard fashion. However, organizations should be cautious not to limit flexibility or ignore user requirements that may be different from those encountered in traditional BI user communities. They must also implement technology that can integrate data from more nonrelational sources, particularly spreadsheets and flat files.

Organizations should keep in mind that giving operational workers wider access to data could have an impact on IT performance; thus, they should make plans for how to keep data flow apace. And wherever it is located, the data itself must be accurate and of high quality; therefore, they should consider creating metrics to assess the quality of the data before it is made available to operational workers. Finally, organizations should prepare to update data delivered to operational workers at least once a day: Nearly 80 percent of participants said that at least that frequency is necessary. As they evaluate technology, organizations should look for software and systems that can support a mixture of query and reporting types and workloads and be capable of scaling up to provide data access and updates as often as necessary.

About the Author

David Stodder is a Research Director for Ventana Research.