Business intelligence (BI) is often seen as a combined business and IT function. As such, the operational structure of BI is not always clearly defined. In this paper, we present an overview of typical BI operational structures and elaborate their characteristics. Based on this analysis, we will (in future work) design a reference process map for BI which can be used as a template to define organization-specific BI operational structures.
Business intelligence (BI) is a topic widely discussed in IS literature and has, as a
combined business and IT function, become an essential component of the information supply
infrastructure and a contributor (and prerequisite) to overall organizational success
In research and practice, various process models (PMs) for structuring the BI function
have been proposed. Some of these models, e.g. the biTIL model by Steria Mummert
Consulting, are based on reference models for IT service management, other models are
more business oriented. But currently, there is no common understanding which
processes a BI organization should cover. Furthermore, it can be doubted that purely
business- or IT-driven models reflect the business-IT-duality inherent to BI.
Therefore, our research goal is to analyze BI PMs currently used in practice and
synthesize a BI reference process map (BI RPM) from these models, which covers all important
aspects of BI and which can be used as a template to define organization-specific BI
operational structures. With the (synthesized) design of the BI RPM, we aim at the
construction of a better IS-related problem solution
Neutral representation Timeframe
Not restriction Organization
Historically
Coverage strategy Selective, as we only want to cover PMs from current practice
In order to document the state of the art of existing BI PMs and to identify their key
characteristics, we conducted a literature review following two established frameworks
Description BI PMs Analyze existing models (integration/generalization) Specialized BI scholars and BI practitioners As we want to analyze BI PMs from practice, we searched the data bases of wellestablished European BI practitioner events (http://awf.unisg.ch/, https://www.tdwi.eu/nc/veranstaltungen/konferenzen/, http://www.infotage.cubeware.de/) for suitable content.
We examine at least titles and abstracts in order to limit the amount of identified literature.
Identification of the key characteristics of the models. For the content analysis and the
synthesis of the PMs, we use a concept-centric approach based on concept matrices
Conceptual models represent selected phenomena in some domain and play an important
role in documenting best-practice data and PMs
BI was initially coined as a collective term for data analysis tools (Anandarajan,
Anandarajan, & Srinivasan, 2004). Meanwhile, BI broadly encompasses all components of an
integrated decision support infrastructure
As there is no universally accepted BI definition
In the following, an overview of four BI PMs from practitioners is presented. The findings of this practice-based analysis are used for gap identification and will be compiled to a list of practice-based requirements for the design of a BI RPM.
The first PM by the Business Application Research Center (BARC)
In order to analyze the content of the BI PMs, we examined the sub-processes in several discussion sessions. We used a bottom-up approach because only the less generic subprocesses allow for comparison. As such an examination is subjective by nature, a standardized process for content analysis and synthesis (conceptualization, codebook creation, coding, refinement, & reliability check) was used, thereby helping to ensure the necessary rigor (Neuendorf, 2002, pp. 50-51). Figure 1 briefly summarizes the results of this analysis, which serve as foundation for the design of a BI RPM. The identified processes are already pre-structured in a way which we believe could be suitable for a BI RPM, but this is subject to future work.
Current BI practices explicitly encompass a rich set of management activities (planning, implementation, and control). BI strategy and governance aspects are good examples for this.
Analysis showed that process granularity is often inconsistent in models. But processes of the same level are required to be of similar granularity. In addition, main process categories should be unambiguous, such that there is as less overlapping between sub-processes as possible.
Three out of four models included comprehensive data management functionality, which makes sense as data is the prime source for BI services.
A generic reference process framework, e.g. ITIL, could improve understandability and applicability of a BI RPM We will use these findings in conjunction with further requirements from academia to develop a theoretically sound and in regards to content comprehensive BI RPM. The result will provide guidelines for developing organization-specific BI operational structures.