Data is the DNA of a manufacturing company. Its quality is crucial in order to make valid statements and well-founded decisions. Regular data cleansing is an important tool for this. Those who use company-wide data quality management to ensure that master and transaction data is consistently generated correctly and processed error-free are laying the foundations for efficient data analysis and value-added services and securing their competitiveness.

Machines, industrial trucks, robots - manufacturing companies invest in hundreds of assets to ensure product quality and lean processes. Too rarely, however, do they focus on the most important thing in the company: the data. This data is sometimes used for years and is the link across all business processes. But if data is not "maintained", it will eventually wear out - gradually and often unnoticed. The consequences are far-reaching and costly: delays in production and delivery, incorrect inventory management, inaccurate production planning or imprecise business decisions, to name but a few.

Business intelligence only works with clean data

In order to avoid these impairments, it is necessary to create an awareness of data quality and data as corporate DNA. Only when management is aware that business intelligence only works with "clean data" can the strategic goal of clean data management be successfully implemented in the company. This increases the accuracy of well-founded decisions because data from different data sources can be checked in combination with a high level of transparency via sales cockpits, dashboards and reports, thus ensuring their quality.

In addition to management, all employees must also be made aware of the potential of high data quality. Specific objectives for data quality management support, for example, more stable customer relationships or greater transparency. All employees must develop an awareness of the impact that supposedly minor inaccuracies in data collection have on costs and their work. For this reason, "business intelligence" starts at the logistics level.

In order to comprehensively assess existing databases, the specialist departments in which the data is generated must also be involved in the analysis alongside the IT department. After all, the users know their data better than anyone else. In an initial data check, ERP manufacturers such as ProAlpha produce a report that documents the data quality (DQ) key figures for the data provided to customers and provides information on how data quality can be improved. At the start of every data quality project, it is also advisable to go through the available data sets with those responsible using specific questions.

Know the status quo of your own data quality

A crucial starting point for data cleansing is knowledge of the actual data quality. It serves as the basis for upcoming milestones such as the feasibility of desired key figures or the preparation of a migration. There are effective analysis tools and procedures for recording and mapping the current status. This makes it possible to identify existing problems with a manageable amount of effort and to identify the frequency of errors. As a rule, redundant master data, incomplete data records and incorrectly recorded data come to light, as well as inconsistencies between different data sets. For example, a compatibility analysis for the new structure of a system is essential to prepare a data migration. This analysis allows incompatibilities in formats and value lengths to be identified immediately and information to be added in advance or data objects to be analyzed that do not need to be transferred to the new system.

The resulting results report is an important indicator for an upcoming data quality project. It can be used as a basis for developing rules that provide clear guidelines with regard to data relevance and the completeness of a data set. With a company-wide glossary, the content and metrics of the data content can be described and made comprehensible for all employees. These characteristics can be used to evaluate the data set and define what should happen with incorrect or incomplete data.

One-off data tuning is not enough

Once the treasure trove of data has been cleansed and polished, management can rely on reliable information from the ERP system and react accordingly. However, the data cleansing project must not be a one-off affair - "keep it clean" is the key to long-term success. In order to ensure flawless data quality in the long term, regular quality controls and data cleansing must be established. The basis for this are flexible and demand-oriented sets of rules with which errors are detected and transferred to the target system, in most cases the ERP system. Employees can then cleanse the data directly.

How often the databases should be checked depends heavily on the respective company and its processes. However, the continuous and regular review and measurement of progress is crucial. This gives companies two advantages at once: On the one hand, good data quality management ensures greater production and process reliability internally, thereby ensuring well-founded decisions. On the other hand, reliable statements about availability and delivery dates increase customer satisfaction and promote cooperation.

Gunnar Schug, Managing Director of the Advanced Analytics Business Unit, ProAlpha