Method

We use a research method inspired from design science, where we construct and evaluate relevant artefacts to enable managerial decision making. This method relies on iterative improvements to the research based on three cycles: relevance, rigor, and design.

Relevance cycle: The research was conducted through collaboration with a medium-sized organization in the retail sector. Through interviews with managers and observations of their data management and decision making processes, we developed familiarity with the current processes and methods allowing us to better define and scoped the problem areas.

Rigor cycle: Literature was reviewed to find existing research results on decision making with an emphasis on approaches based on BI and goal and process modelling. This step helped us to better understand the state of the art and related research and to validate our research topic.

Design cycle: We built model artefacts of a real-life example to motivate, illustrate and evaluate the main aspects of our decision-making methodology. Feedback from real users was incorporated in our methodology during its refinement. Tool support was developed to demonstrate the feasibility of the methodology. Lessons learned from the design cycle were also documented to augment the knowledge base related to goal-oriented and BI-driven decision making.

The main contributions of this paper include:

A novel goal-oriented decision methodology that integrates goal models, decision models, threats and opportunities (collectively called situations in this paper), and processes together with analytic capabilities in order to improve decision-making capabilities within BI systems.

Extensions of a standard goal-oriented language (from the User Requirements Notation) to better model and analyze relationships between KPIs, goals, and situations. In particular, these extensions enhance what-if strategy analysis in support of decision making.

Tool support for the above extensions and for integration of goal/decision models with BI systems (particularly with a commercially available system, namely IBM Cognos BI).

The rest of the paper presents our literature review (resulting from the relevance and rigor cycles), the methodology that exploits goal and process modelling together with BI for decision making (from the design cycle), and the application of the methodology to a real-life example from the retail organization (also from the design cycle).