Produce good BI results
The Business Intelligence architecture will not be able to produce good results if we cannot link tools to it. This blog provides an overview of the most important Business Intelligence tools (BI Tools) in relation to the decision-making cycle. It also describes the purposes, the features and the workings of these tools.
The decision making process
Each Business Intelligence Tool plays a supporting role in one or more phases of the decision-making cycle. The decision-making cycle comprises of the following phases, which can be supported by the following tools (see figure below):
Collecting information using dashboards and reports
The decision-making cycle starts with gathering the right information: ‘scanning’ the environment (or the internal organization) in order to see if a problem exists. Mainly dashboards and reports are used for this purpose. This first phase should be seen as a ‘wake-up’ moment during which any current or potential deviation from the desired situation is a decision problem. In order to qualify or trace something as a problem, the first phase should – in principle – include (or be preceded by) a process in which we identify our organizational objectives.
Designing and analyzing using interactive analysis and data mining
Once a problem is discovered, we know where the issue resides but we do not yet know its exact location or its whys and wherefores. For example: the customer satisfaction indicator on the dashboard flashes. Subsequently, we will want to find out what causes this. We may, for example, use interactive analysis ‘on-the-fly’ to split out customer satisfaction per region, per account manager and, if necessary, per product group.
We might also use interactive analysis (OLAP) in order to easily compare the number of complaints with the customer satisfaction in the same region and period. By doing this, we gain insight into the reasons why customers are less satisfied – a recently launched product exhibiting start-up problems, to name a random example. We use data mining – discovering ‘hidden’ relationships – to gain insight into characteristics that are a determining factor for more complex issues such as fraud. In such cases, interactive analysis is not sufficient because we generally lack the required detail data. In addition, the complexity associated with discovering such (causal) relationships is simply a bridge too far for interactive analysis.
Very crucial for strategic decision making
“This phase in the decision-making process seems to be crucial for strategic decision-making, because it is during this phase that we determine our entire course in terms of decision-making.” (Mintzberg, 2004). It is thus important to apply these tools well and to adjust the organization accordingly. This also means that an effective working relationship should arise between managers and analysts so that the intuition and the authority of the managers can go together with the analytical brain of the analyst and with the methods and functionalities available to him (or her).
Selecting and implementing using ad hoc query, what-if and forecasting
Once we have established the cause of a problem and we have mapped the (causal) relationships, we can take action – in consultation with managers and stakeholders. Ad hoc query (‘ad hoc’ querying of data sets) can support this. We may for example use ad hoc query to create a detailed set of customers whom did not place any orders in the last month, based on the results of the interactive analysis. Subsequently, we can write a letter to the selected customers in which we emphasize the benefits of doing business with us and include some interesting offers. We use ‘what-if’ and forecasting in order to calculate actions in advance. In this way, we gain insight into the impact of our actions before we actually implement them.
Evaluating using dashboards and reports
Once we addressed a certain problem by taking action, we use either a dashboard or reports in order to assess whether the problem has indeed been resolved. It may be that the problem reflects a key success factor of the organization, which we did not know about yet. If this is the case, we can place the key success factor – including the associated KPI – on the dashboard or in the report so to permanently evaluate it. In this way, it becomes an integral part of the organization’s management model.
Separate components that are hardly related
Relatively many organizations still appear to regard the above-mentioned tools as separate components that are hardly related. For Business Intelligence to be effective, it is however, of the utmost importance that the tools can work together very closely so creating a solid and smoothly operating entity. Users should ideally be able to both use and apply the tools more or less unwittingly in their daily activities. Metadata, web technology and a proper architecture play an important role in achieving this.
Increase the coherence between the tools
Additionally, the suppliers of BI tools undertake actions – based on the integral needs expressed by users – to increase the coherence between different tools. Portals are a typical example of this development. However, we cannot start celebrating yet: practice shows that when it comes to integration of different tools within the portal, there is still a lot to be desired. Unfortunately, it also still happens that certain functionalities are present in the desktop version , but have not yet been implemented in the web version.