Dirty Data
The problem of “dirty data” refers to virtually all Business Intelligence implementations, regardless of the size of the data. In every natural environment, every computer system, there are errors and omissions in the data. This includes the problem of so-called dark data, or data which is not used, both due to the fact that it is useless, and because it is poorly organized and therefore cannot be effectively processed. It is therefore important to devote sufficient time to eliminating inconsistencies and to determining their origins, as well as to the arrangement of the data that is valuable in the analytical process. The “cleansing” of the data is very important, because all incorrect or incomplete information will affect the quality or even the correctness of the results of reports and analyses in the BI system.
Too much data
The amount of data stored and generated daily by companies is very large and constantly growing. However, this doesn’t mean that all this data will be useful for the implementation of the BI system. A difficult but necessary task is the selection of appropriate data – even at the stage of pre-implementation analysis. The integration of other data will result in prolonging the analysis and implementation processes unnecessarily (and thus also increase the cost), but does not bring any real analytical or business benefits.
Nobody to take responsibility for the processes and data
During the pre-implementation analysis it often turns out that there is nobody who is clearly responsible for some of the key processes. This just goes to extend and complicate the analysis and the subsequent implementation of the system, due to the fact that these “orphan” areas are very difficult to verify, clean and implement.
Differences in data during migration
During the implementation process, within which the migration of a current solution to another tool takes place, it often turns out that the data in the new system is different in relation to the previously used solution. This is a fairly typical situation and results from changes made to the system during use, and as a result the guidelines connected with the starting point and adopted for the migration may differ from the existing condition of the system. In such situations, cooperation between the developer and end-users is vital to identify the source of the problem and make it clear that the new solution is trustworthy.
Lack of openness to change
Even the best system would barely be worth having if there were nobody to use it. A Business Intelligence system must be based on real needs and the actual interest of potential end-users within the company. In a situation where employees are resistant to change, it is essential to ensure that they come to trust the new tools, and that in turn their needs and objectives are met in full.
Read also: How Not to Come Unstuck When Implementing a Business Intelligence System – Part 2
Business Intelligence Consultant, who design and implements Business Intelligence solutions in Microsoft SQL Server, using both standard tools and Power BI functionalities. He has experience in workin wirh Oracle, SAP, IMP, QlikView and Pervasive.