The chapters 2 to 6 presented Business Intelligence architecture scenarios. You could see the different layers, when they are used when not. What was always the last layer in these scenarios, was the application layer. This layer is the access point for end user or business users in order to access the data and generating information out of it. If you “google” business intelligence applications, frontends or something similar, you get a lot of websites with descriptions, explanations and definitions. But at the end, you are more confused than enlightened.
So, in the next basics, I will present all different application classes. It will start with dashboards and ends with legal consolidation. These application classes allow different types of designing data for providing insights and creating information. The application layer is responsible for the visualization of the data through tables and charts in different ways. The data is provided in a way end users can work with it based on their skills and requirements. They are responsible for the user satisfaction and business advantages but are only a small part of the whole business intelligence environment efforts. In general, these business intelligence (front end) tools are chosen by the IT-department or business department. That leads often to problems in the business intelligence architecture, organisational aspects and to complications between the different departments. The best way is to evaluate a short list of tools based on the wishes and requirements of the both sides.
In my former job as analyst at BARC (a le CXP company and the European leading analyst for business intelligence), we used a two axis chart to categorize the different applications. What you see on the chart is the x-axis with the complexity which rises from left to right. On the y-axis, you see the freedom in usage increasing from bottom to top. In the center you see the different application classes:
- Standard Reporting
- Ad hoc Reporting
- Planning & Simulation
- Data Mining
- Legal Consolidation
Dashboarding tools provide a high flexibilty for creating applications with high density of key performance indicators (kpis) which are consumed via the computer screen. Standard Reporting means the creation and distribution of predefined and/ or precalculated reports in regular periods. The end user can only consume the data, but cannot change anything. Ad hoc Reporting is the possibilty for business users to create queries and reports or dashboards with content. Analysis provides a flexible navigation in the data and/ or the usage of predefined analytical methods. Planning is the creation of new data, simluation is the calculation of measures based on the existing data. Data Mining is data analysis based on statistical methods. Legal consolidation is the creation of one view on the company based on the accounting. The outcome of a legal consolidation can be a balance sheet and a profit and loss statement.
On the market for business intelligence tools, you can find generalists which have tools for more than one application class as well as niche players which only provide software for one application class. While Tableau and Qlik adress primarily analysis, does SAP or another company of the big FIVE (IBM, Oracle, Microsoft, SAS and SAP) have more tools for more than one application class in their product portfolio.
Dashboarding tools provide a high flexibilty for creating applications with high density of key performance indicators (kpis) which are consumed via the computer screen. Dashboard applications like balanced scorecards provide functions, processes and visualizations to implement the balanced scorecard philosophy.
I do not want to say that these dashboards or balanced scorecards are good or bad. I want to show that a typical dashboard or report is a one pager with all relevant figures on a screen the end user needs to do his daily, weekly or monthly decision.
Balanced scorecards or dashboards are developed with business intelligence development environment. These tools allow to create individual information systems for individual user groups. These development environments are used by it-driven users to create these applications. End users like managers only use the applications.
Standard reporting is the creation and distribution of predefined and recalculated reports. The reports will be sent automatically at a specific time and date period and can be daily, weekly, monthly, quarterly or yearly. The report has a specified layout and will be filled with data when the report will be proceed by the system. The report recipient gets the data and can consume it, but is not allowed to do any changes. Standard reports are made for a static reporting when data is needed in regular time periods.
Tools for standard reporting need a full data access, high layout control and perfect possibilities for formatting tables and charts for the developers. The created report will be send via e-mail or distributed via a portal and can be an HTML-, PDF-, or Excel-file. In a portal, users can open the reports and do have the possibility to export or print them.
The typical standard report approach is a two-user role concept: report developer create reports for end users who can consume the provided data.
Ad hoc Reporting
Next is the application class ad hoc reporting. It is the class with the more complexity and freedom at the same time for end users compared to dashboarding or standard reporting, but requires more skills.
End user in the business departments can do queries and create reports with individual data if they do ad hoc reporting. This step requires knowledge about the data, the way of execute queries and other skills. It is possible to reduce the complexity of the data sources and the access to it by creating a semantic layer which protects the business user. The semantic layer is a easy-to-use platform for low skilled business users.
The semantic layer is harder to show in a real picture. What you see in this picture are several source systems and a semantic layer which is named as “data model”. This is the access point for every end user tool.
Ad hoc reporting does have three role model. Developers develop and define the data access and the data model in the semantic layer which can be understood by the end users. Power users in the business departments use the semantic layer to create dashboards, reports and simple analysis. Report recipients consume the created applications.
This class has three main forms.
Tools for analysis provide functions for a flexible navigation in data. In general, these tools come with standard analysis methods like an ABC analysis. Advanced analysis tools provide mathematics or statistical functions and easy-to-use methods for data mining. The main focus is still on the analysis on the data, not on developing complex data mining algorithms.
There are three main forms of analysis:
Dimensional analyzes allow a flexible navigation in predefined, modelled dimensional data. There are three forms of dimensional data (ROLAP, MOLAP, HOLAP but will be explained in a following chapter). There are special funtioncs like drill down, roll up, slice and dice or drill through which allow the user to browse through the data. Typical terms are hierarchy and dimension.
Associative analyzes allow to analyse objects (e.g. products, customers), their elements and their relations. The data model of these analyzes does not hierarchies or dimensions because they are not part of the data model. Instead, relations between the data can be analyzed.
Visual analysis users use basic and advanced chart types in order to analyse data. Basic chart types are lines, pies and bars (there are several thought leaders who recommend the waiver of pies). Advanced chart types are for example scatter plots, bubble charts, needles, maps, treemaps or heatmaps. Visual analysis does address more users compared to dimensional and associative analysis because it is more intuitive. But to analyse data perfectly with charts, a lot of education is needed about the possibilities and threats of specific chart types.
In general, there are no user groups, but one person which does the analysis. This person may create his/ her own data model sometimes. The difference between analysis and standard and ad-hoc reporting is a information creation. While the combination of analysis methods and data creates information. Tools for standard or ad-hoc reporting only visualize the provided data.
Planning and Simulation
Next is Planning and Simulation. This class means a new chapter because it alllows the creation of new data through the user.
Planning means the creation of data with a future time reference. Simulation is the calculation of values based on actual data and calculation methods.
The main task of planning software is the support of planning user when they want to plan their business with planning formulas and distribution algorithms (e.g. seasonal distribution, distribtution of yearly budgets on months). Planning processes can simplified via planning procees controls (e.g. release, closing, disclosure).
- IBM TM1
Planning and Simulation does have one person at least. In general, there are more than one person involved in planning processes and simulations. Planning processes can be done top down (management plans the budget for a whole group), bottom up (several persons plan the budget for their segment/ area) or mixed planning (management plans, level below distributes the figures).
Data Mining means the computer based data analysis with statistical algorithms. These algorithms search for patterns, exceptions and abnormalities in the data. The possible outcome can be a segmentation, classification or association of data. Methods can be statistical functions, neural network, decision trees and other algorithms.
Application: IBM SPSS
User of these tools have to highly educated in order to use these software and the algorithms and to validate the results. The integration and preparation of data is so important for the user that data mining tools bring addtional functionality for these tasks.
The last application class is called “Consolidation”. This class is not a typical application class because the freedom for the user is low but the complexity is high. This is a result out of the specifications.
Consolidation means the creation of one view on the business. In practice, there are two forms who are highly relevant:
- Legal Consolidation
Legal Consolidation means the creation of one view on the business based on accountancy. Legal consolidation observes all legal specifications. The profit and loss statement as well as the balance sheet are compulsory.
- Management Consolidation
Management Consolidation means the creation of one view on the business based on accountancy. The difference to Legal Consolidation is that profit and loss statements as well as balance sheets are only done on the management specifications. The figures in these applications can different from the legal`s.
Legal or Management Consolidation is done by consolidation experts which have a deep knowledge in the business organization and legal specifications. In general, consolidation software supports the users with predefined consolidation models based on the legal specifications for each country.
About Tobias Riedner
Tobias Riedner foundet WYCDWD.com in 2015. He works and worked as innovator, consultant, analyst and educator in the fields of business intelligence and data warehousing. He learned a lot from the best consultants, managers und educators in the past and shares his knowledge worldwide. He works for a steady growing traditional company which is a leader in industry 4.0.