According to the Gartner Group, over 70% of Business Intelligence (BI) projects fail. Among the reasons are the di erent languages employed by IT and business people and the necessity of a long-term BI plan describing the goals of the organization. In current practice, when building the data warehouse (DW), strategic plans are rarely considered. In this paper, we propose a method to align the business plan with DW requirements analysis. By aligning the DW, we (i) validate the correctness of each decision makers' goals, (ii) ensure that their decisions and the DW contribute towards organization goals, and (iii) provide a longterm uni ed BI strategy. We instantiate this alignment by combining i* for DWs with strategic business models.
Data Warehouses (DW) integrate numerous heterogeneous data sources in
multidimensional structures in support of the decision-making process. In order to
be successful, recent DW development approaches, such as [
The main drawback of the techniques used during the requirements step is that they obtain only a partial view of the problem from each decision maker. Moreover, these partial views may not always be aligned with the business plan, which, together with the gap between IT and business people, makes it di cult to validate the goal models. In turn, the lack of alignment between individual decision makers and the business plan is translated into a lack of long term enterprise BI strategy, which is one of the key factors to successfully apply BI. This problem arises whenever business goals are not considered during the requirements analysis step. For example, a ctitious car dealer company EURent pursues the goal of \Be positioned within the top 10 car dealers". EURent Sales Manager is considering how to improve the \Car Sales" process by \Increasing the margin of bene t per sale". While this particular goal improves the results of the sales process, it does not contribute to the increasing number of cars sold.
In this paper, we propose to tackle this problem by aligning the business strategy with current BI and DW goal models. Therefore, we ensure that our BI-enabled decision making is consistent with the business strategy. First, we elaborate a strategic goal model from the business plan to formalize business goals and trace business indicators. Then, we align and relate decision makers' goals with business goals. Therefore, we ensure that all the decision makers are contributing towards the overall goals of the enterprise, and identify business goals that are being overlooked by decision makers.
The remainder of this paper is structured as follows: Section 2 presents the related work in DW requirements. Section 3 describes the alignment process and the main contributions of our work. Section 4 presents the conclusions. Section 5 describes the ongoing and future work in this area. 2
Until now, most of the attention in DW requirements analysis has been focused
on requirements elicitation. Current techniques [2, 5{7] are focused on
gathering requirements by means of interviews and questionnaires. However, although
it has been stated that DW requirements can rarely be gathered correctly and
comprehensively from individual decision makers [
In this section we present the main contributions of our work, as well as the
current tool support for our approach. Our contributions are described in the
following order: (i) a method to perform the alignment between DW requirements
and business goals, (ii) a set of mappings to instantiate the alignment method,
by using i* for DWs [
First, the alignment method is shown in Figure 1. The alignment process starts by modeling decision makers' goals by means of techniques used in current DW approaches. As a result, an i* for DWs model is created for each decision maker. In parallel, or after DW requirements have been obtained, the business strategy model is created by using the information from the business plan. Business plans include information regarding business goals, the objectives associated with each goal, and may also include information regarding business processes. Once both models have been obtained, we proceed to the alignment step. First, we align the concepts used to model decision makers' goals with those used to model business goals, in order to ensure that we correctly relate individual goals with business goals. Then, in collaboration with a domain expert, we align each concrete decision maker goal to a speci c business goal. Finally, we analyze the alignment and perform changes as necessary.
Since currently there is no standard for modeling decision makers' goals nor business strategies, they can be instantiated according to di erent frameworks, thus leading to di erent results. Therefore, in order to instantiate the method, it is necessary to perform an ontological mapping between the concepts of the speci c frameworks being used. In our case, we make use of i* for DWs in order to model decision makers' goals, and BIM for modeling the business strategy. Thus, we analyze the de nitions and characteristics of each concept included in these metamodels and align them to ensure their correct use. Due to space constraints, we provide only a brief overview of the metamodels, and focus speci cally on aligned elements.
First, i* for DWs [
Second, in order to create the business strategy model, we use BIM [
According to the descriptions provided in the i* for DWs and BIM proposals, Decision and Information Goals, and Contexts from i* for DWs cannot be aligned with any element in BIM. The main reason is that BIM focuses on representing the business strategy and does not consider elements speci c to the individual decision making process, such information requirements or actors.
Next, Business Processes and Measures from i* for DWs can be easily aligned with BIM elements. First, Business Processes refer to the same concept as Processes. Second, Measures provide a value vi that measures the performance of a business process. These values are one of the basic elements used to calculate Indicators that monitor elements in the business strategy. A Measure can be related or transformed into an Indicator Ij if (i) vi is used in the calculus of Ij and (ii) the rest of the values tj , thj , and wj are de ned.
Finally, Strategic Goals can be aligned with Intentions as follows: according to our formalization, achieving a particular decision maker strategic goal Strategic Goal Strategic Intention
Future, Focused,
Long-term
Qualitative,
Measured by KPI Future, Time-Targeted,
Long or Short-term, Measures Goal and
Process performance Si improves a process BPm. Thus, this improvement helps the organization to achieve a business intention Ij , realized by BPm. There are, however, two limitations to automate this alignment. First, Si may improve a speci c objective Ok that does not have an impact on Ij . For example, improving the \Car Sales" process by \Increasing the margin of bene t" does not have an impact on \Be positioned within the top 10 Car Dealers". Second, in practice, not every business plan or strategy provides references to the business processes that realize each intention. Thus, it is necessary to collaborate with a domain expert in order to identify if the particular goals are aligned with the business intentions or not.
The alignment is summarized in in Table 1, along with the informal descriptions provided by the frameworks for each element.
After aligning the concepts, the last step is aligning the elements within the goal models obtained from decision makers with the business strategy, thus obtaining a combined model that covers both individual as well as organizational views. We have successfully tested our approach in an experimental case study, creating a business strategy from a business plan and interviewing experts in information analysis to obtain decision makers' goals. However, due to space constraints, the case study will be described in a future extended version.
Our approach is supported by our tool, the Lucentia BI suite, which allows
us to model (i) user requirements by means of i* for DWs, (ii) business strategies
by means of a particular implementation of BIM, and (iii) a trace metamodel,
described in [
In this paper we have presented an alignment between the business strategy, modeled using BIM, and DW requirements, represented by i* for DWs. Our process results in an alignment that allows us to (i) validate DW requirements according to the business strategy and identify non-aligned goals, (ii) provide a long-term BI strategy to be pursued, including what information is being used by the organization to support each goal, (iii) identify the di erent decision makers participating in a business goal, thus providing awareness, and (iv) evaluate if decision makers are being successful by analyzing the values of indicators related to business goals.
While extending i* to consider all the elements within the business strategy would overcomplicate the model, our initial applications have shown that it can be combined with other models in order to capture both the particular viewpoint of each decision maker as well as the overall strategy of the organization. 5
The current ongoing work is focused on making it easier for businesses to apply our proposal. In order to achieve this, we plan to semi-automate the process of obtaining an strategic model directly from the business plan. Therefore, we are analyzing the viability of de ning a series of pattern-based transformations in order to save time and costs.
In the medium-term we plan to apply our approach to a real case study and evaluate the results. Since then structure of business plans may vary from one organization to another, we plan to minimize the impact of this variability by using the standard Business Motivation Model, proposed by the Object Management Group group.
Acknowledgments This work has been supported by the MESOLAP (TIN201014860) and GEODAS-BI (TIN2012-37493-C03-03) projects from the Spanish Ministry of Education and the Spanish Ministry of Economy and Competitivity. Alejandro Mate is funded by the Generalitat Valenciana (ACIF/2010/298).