Inception of the PhD project idea

As mentioned earlier, the transaction systems alone produce 9 million transactions per day. This results in a massive amount of data available in the systems of the bank. In the 1990s the bank started to use early BI systems and practices by running analyses on historical data to observe patterns from which new insights might emerge and lead to better decisions. As one of the interviewees noted:

I think it was trial and error or finding out yourself but also the tools we used at that time, someone told you, oh, this is the syntax here and that's how you do it. So the introduction was 30 seconds, and then trial and error I guess. But this is really many years ago, but that is how I record it, you know, that no courses, no big books, nothing. That is somewhat a mass of data made available to me as business user, in a structured way and I can access them, I can make out of those my own reporting, own analysis. (IT Finance Business Analyst)

At that point in time the BI practices were scattered amongst the different departments and areas. Many times, each of the areas would create their own databases with relevant data and analyze those using their own BI tools.

Today, in Group IT, there is a data warehouse department in charge of systematically gathering raw data from the source systems, including all the different transaction and administrative systems of the bank, and storing these data in the data warehouse so that they are accessible to organizational members who can in turn retrieve data for further analysis. The data are retrieved through different BI tools such as query analyzer or OLAP tools, which allows users to analyze multidimensional data from multiple perspectives. However, until 2008, while there was a data warehouse department, it included many different warehouses with some areas having their own, rather than a single, consolidated warehouse. Further, in these data warehouses it was often not very easy to perform ‘slicing and dicing’ of data according to particular user needs. In addition, there was not a standardized data model used by the source systems that input the data to the data warehouses. This created a proliferation of the duplicate entities with different names, i.e. different systems used many different names for the user ID, making the integration of data from different domains very difficult.

In 2008, Group IT decided to invest in a new BI platform that would incorporate state of the art technologies, processes and methods. The aim was to create a more resilient data warehouse, common data modelling practices, a common repository and BI tools, and to make these accessible to decision makers all over the group with the ultimate goal of making better decisions. The idea of combining, integrating and sharing data between and across different business domains was crucial to the new initiative. A new department was formed, the BICC, to carry out the development of the new platform.

At the same time, in the spring of 2008, the management of Group IT announced to the development areas and their departments that a new Industrial PhD project would be financed that year and that they should write proposals if they were interested in

conducting research in their areas. The chief architect of the BICC wrote a proposal with the title: Cross–domain alignment, traceability and precision management. The purpose of the project was to support and improve the decision-making process by developing a framework to integrate and align business intelligence from different domains and existing IT applications.

The management of Group IT selected the above proposal among others. They felt that many decisions, especially in the Group IT itself, were based too much on “gut feeling” rather than data and analysis. They felt that there was room for improvement in their decision making processes. Characteristically, in one of the first meetings with the CIO he concluded by saying:

The point is that we need more data to make our decisions! (CIO)

With the above proposal as a starting point, and before deciding on which decision-making processes I would focus on, I began a pilot study on how BI was understood by and used in the organization. During this pilot study I interviewed people from the following areas: Group IT, Group Credits, Group Finance and the Danske Bank DK brand (regions and branches). I started with Group IT, as I was an employee there, to investigate how BI was understood and how decision-makers used it. From there I identified analysts and decision-makers who used BI applications in Group Credit and Group Finance. Group IT has been already introduced, but below I provide a brief introduction of these other areas in which I conducted the interviews for the pilot study.

The Group Credit is a cross-organizational unit responsible for credit policies and reporting and monitoring the credit portfolio. Further, the unit is responsible for risk management and pricing across business units and customer segments. This group uses data analysis primarily to develop the credit systems that calculate the risk profiles of customers, to approve or decline loan applications, to enhance the valuation systems and to adopt price fees for loan applications. The credit process is one of the main

users of data from the data warehouse. Several BI applications and standard reports have been created for the Group Credit.

Group Finance is another resource area that is responsible for performance management, accounting and investor relations as well as law and compliance across the whole Danske Bank Group. The performance management activities concern data gathering and analysis to track the performance of all brands and resource areas. These performance management activities are supported by a BI application called the Group Management Information (GMI) system. As one interviewee reported, this

… is a business intelligence system primarily targeted for the branch network to enable them to make the decision locally in the market based on the best foundation. (Performance Management Specialist)

Since the GMI system was also used in Danske Bank DK (one of the brands, see Figure 7), I further explored the use of the BI output in decision-making in the regional units of Danske Bank DK and its branches. Danske Bank DK, as the brand that accounts for almost 50% of the earnings of the group, offers retail services to customers in Denmark. The headquarters of Danske Bank DK perform various analyses to track the performance of its branch network, which consists of regions and each region consists of several branches. Regional groups do the same for their own network of branches and the branches for the performance of their employees (advisors). The main BI application they use to perform analyses is the GMI system and the query analyzer to extract and run SQL queries.

During the pilot study I realized that when referring to BI decision makers would refer to the technologies, tools or applications or the process of accessing, retrieving and analysing data. However, when referring to the use of BI in their decision-making practices it was clear that BI was not used in organizational decision making as a process or a technology, rather it was the output of the BI process and of BI technologies that was used in decision-making. That is, in decision-making BI was

perceived by the makers as an output/product of a process. The decision-makers at the bank would refer to this output as ‘the outcome of the analysis’. As already defined in the introduction, the BI output is the outcome of data-driven analysis that decision makers use in their decision practices. Further, as already mentioned in the introduction, decision makers found it easy to talk about the process and technologies used to generate BI output but difficult to elaborate on the use of that output in decision practices.

This realization, in combination with the findings of the literature review, made me change the research scope. Instead of focusing on improving the decision-making process by developing a framework to integrate and align business intelligence from different domains and existing IT applications, I decided to investigate the role of BI in making and specifically how the BI output is used in organizational decision-making processes. This decision was made because there was a need both in the organization as well as in the BI literature to understand how the output of BI is used in organizational decision-making processes.

The next step was to decide which organizational process or processes would be included in the investigation. I chose to investigate a process of strategic importance to the organization: the process of IT project prioritization. This decision process is characterized by high complexity, uncertainty, ambiguity and it is inherently political.

These characteristics make this decision process more interesting as a setting to investigate the role of BI, which is much more obscure in such a process.

In the next section, I provide a detailed description of the IT project prioritization process in the organization and of the BI output used in this process.