BI projects are often accompanied by special requirements and therefore require a special approach. In this article, you will learn how to take these into account in project management and master them with the help of clearly defined success factors for BI projects.

Many BI projects fail – up to 85% of all big data projects, according to Gartner. The causes are often avoidable. To become part of the successful 15%, we present you with tried and tested key principles for successful BI projects, which are often underestimated in practice.

Where do we get our experience from?

29FORWARD has already successfully completed a large number of BI projects worldwide in the banking, insurance, retail and aviation sectors. As a result, we have built up wide-ranging expertise in BI project management and are familiar with the special dynamics of such projects.

Data quality and technical significance of KPIs (Key Performance Indicators)

In some companies, those responsible already recognize the strategic importance of data and its quality. Others, however, do not yet have a clearly defined concept for this. Clear responsibilities, the quality of individual data and the definition of the content of KPIs are among the key success factors for BI projects. Basically, the higher the quality of the data, the more meaningful the KPIs calculated from it will be. Reliable decisions can only be made on this basis if the underlying data is complete, consistent and comprehensible.

Success factors:

1. A clear definition of responsibilities for data and data changes is necessary. It is important for your project team to transparently delineate the responsibilities and interests between the various stakeholders. This also includes the involvement of the line organisation, which requires close cooperation with bodies such as data stewards or a BICC (Business Intelligence Competency Center).
2. Provide the relevant department with tools to measure and improve data quality. You can define your own metrics for data quality, e.g. consistency of a data set, completeness of a data set, redundancy (no duplicates).
3. Ensure that your documentation is up-to-date and complete. Always keep the documentation of your KPIs and data streams up to date and ensure a transparent and standardised understanding within the project team. In this way, you avoid a heterogeneous interpretation of key figures and provide a single point of truth for the relevant company key figures.
4. Motto “Trash in – Trash Out!” – If possible, rectify data quality problems in the supplying systems in the source system to avoid time-consuming data corrections in the subsequent BI system. The golden rule applies: A BI system is only as good as the quality of the data provided in the sources.

Test management

As BI projects are heavily based on data, test management plays a central role – especially the creation of suitable test data in addition to the actual test execution. At the beginning of many projects, however, those involved often underestimate the complexity of this task.

Success factors:

1. Make the specialist departments aware of the importance of testing in IT projects. Software quality can hardly be assessed without extensive testing. To ensure consistent quality and to relieve the testers, the possibility of automated testing should also be examined. This applies in particular to the execution of regression tests.
2. On the other hand, software testing also has clear logical limits. Basically, testing can only prove the presence of errors. This means that a system is not error-free just because all tests were successful. This basic understanding should be internalized by the entire project team in order to be able to plan a realistic test procedure.
3. Experience shows that an incremental approach should be used to break down large work packages into smaller, more manageable packages, particularly in order to maintain a high level of software quality. In this way, errors are identified at an early stage and not just at the end of the project. In addition, lessons learned can be implemented early on in development and testing in order to optimize the further course of the project.
4. To comply with regulatory requirements (e.g. the GDPR), set up the test data with anonymized and pseudonymized data. This procedure should be transparent and coordinated with the responsible contact persons in the organization (e.g. the data protection officer) at an early stage.

Regulatory requirements

In principle, regulatory requirements (GDPR, critical infrastructures, MaRisk, BAIT, etc.) naturally apply to all types of IT projects. Aufgrund der Verarbeitung und Analyse von oft sensiblen oder vertraulichen Daten spielt der Datenschutz in BI-Projekten jedoch eine besondere Rolle.

Regulatory bodies such as auditors or data protection officers are often perceived as “impeding innovation”. Project management urgently needs to avoid such a working environment. It is important to mediate between the various stakeholders in order to constructively bring together the requirements of all parties.

Success factors:

1. Actively involve central roles from data protection, IT governance and compliance at the start of the project. Integrate them into the project organization and communication. Clarify all regulatory requirements at an early stage – taking into account the existing IT infrastructure. Define clear framework conditions for implementation. Record the jointly defined goals in the implementation concepts of the respective work packages. Have these checked and approved by the responsible stakeholders as part of an official acceptance procedure. This will safeguard your project and avoid later conflicts or costly workarounds.
2. Analyze the BI solution and the data budget for sensitive and risky data and functions. Develop an authorization concept for each system component together with stakeholders from governance and IT infrastructure. Base this on the requirements of the GDPR and MaRisk. Have the concept checked and approved as part of an official approval process before implementation.
3. Define and document all processes at an early stage – for example, for anonymization and pseudonymization. Adapt the authorizations to regulatory requirements, the information needs of the specialist departments and risk management. Record all relevant information and requirements as early as the system and architecture design phase. Plan bindingly and with foresight in order to have enough time to coordinate the authorizations with all stakeholders.

Our conclusion on the recipe for success

There is great potential in every BI project, taking into account the success factors mentioned above. Always bear in mind that it is the responsibility of project management in particular to develop the joint concept in close cooperation with the responsible company roles in order to avoid conflicts and unnecessary additional work. You should also involve test management, regulatory affairs and data quality in the conceptualisation phase at an early stage. Using an iterative or even agile approach, BI solutions can be developed step by step. Experience has shown that this is particularly suitable for BI projects in order to best apply our success factors in practice.

The content and success factors in this article are based on the extensive experience gained from 29FORWARD projects. If you are further interested, 29FORWARD is also available to you as a trustworthy partner for the realisation of your BI projects.

Source: [1] Matt Asay (2017): “85% of big data projects fail, but your developers can help yours succeed”, URL: https://www.techrepublic.com/article/85-of-big-data-projects-fail-but-your-developers-can-help-yours-succeed/, Datum des Zugriffs: 23.07.2021