The digital transformation is leading to an explosion in the volume of data. Today's companies must be able to deal with growing volumes of different data and constantly make use of new external and internal data sources. Modern analysis tools provide valuable services here. With advanced analytics, a logical extension of business intelligence (BI) and business performance management (BPM), even future developments and events can be predicted relatively accurately.

While traditional BI tools examine and analyze historical data, advanced analytics is able to automatically identify hidden patterns and correlations in existing data and project them into the future using statistical methods. If the application has an AI (artificial intelligence) module, it even learns to make further forecasts based on the results. This significantly simplifies and accelerates planning processes and increases the level of detail in planning. Advanced analytics applications also make it possible to create what-if analyses - for example, to anticipate the effects of a new business strategy.

More and more organizations want to benefit from these advantages. According to a study by the Business Application Research Center (BARC), ten percent of German companies stated at the beginning of last year that they frequently use advanced analytics. At the beginning of 2018, the figure was five percent. The main drivers are the specialist departments: In almost a third of the companies surveyed (31 percent), key users from the specialist departments work with advanced analytics. However, BARC's analysts assume that the number of occasional users will also increase in the future. This is because not every evaluation is based on highly complex algorithms that have to be modeled and adjusted and require in-depth analytical and statistical knowledge. Employees from the specialist departments can also create simple forecasts themselves after appropriate training, provided they have a good knowledge of the respective business case and a certain understanding of the data.

Meaningful database, defined framework conditions

However, advanced analytics also has its limits. Simply feeding statistical models or artificial intelligence with data will not lead to fundamentally new results in the short term. Certain framework conditions and a realistic assessment are required to be successful here. In order to discover patterns and correlations in existing data and derive valid predictions from it, the basis must first be created. This means that the right data must be selected and integrated in the required quality and granularity and with sufficient history.

A deep understanding of your own business model is also crucial for the successful development of models. This is because measures derived from observed factors can only achieve the desired effect if there is a cause-and-effect relationship between them. Forecasts based on correlations without causality are generally not helpful for corporate management.

Moreover, advanced analytics is not a crystal ball. Predictive algorithms reach their limits when it comes to events that cannot be learned from past data. For models to retain their validity, the relationships depicted must not change in the future - they must be derivable from the past. In the case of dynamic relationships, for example from a volatile business model, it is difficult to use historical data for valid and longer-term forecasts. Extreme values or previously unknown dependencies can also falsify the forecasts. A sufficient historical context that can place the observed factors in a cause-and-effect relationship can only be obtained if the framework conditions do not change frequently.