Data: Probabilistic vs. Deterministic

In order to survive against competitors in the long term and meet the demands of customers, more and more companies are turning to data management platforms or other external data providers. The ever-growing demand for high-quality and meaningful data is forcing industry players to rethink their approach in 2017 if they also want to rely on advanced marketing solutions such as cross-device targeting. 1plusX AG explains.Probabilistic vs. deterministicProbabilistic data, in contrast to its deterministic counterpart, is not based on clearly assignable customer facts that result, for example, from a user login. Instead, the probabilistic approach is based on completely anonymized profiles generated by statistical or machine learning methods. Technology in the field of artificial intelligence has developed so rapidly in recent years that the quality of the data it generates has improved enormously. For many new solutions to optimize ads and content, a large amount of data is needed today, which in most cases can only be collected through modeling techniques. This is also the case with cross-device targeting.Cross-device targetingThere is a big demand for cross-device data on a probabilistic basis. The focus is on connecting desktop and mobile devices and getting the full picture of a potential customer. Companies are gradually realizing the benefits this "big picture" brings them. Depending on the time of day and the end device, content and offers can be optimized and provide the user with a consistently positive customer journey.TV and radio devices are also increasingly being assigned to households and thus also to individuals on the basis of probabilistic calculations. This trend will continue. To achieve sufficient coverage for cross-device targeting, an enormous amount of deterministic data would be required, which is a) costly, b) difficult to generate, and c) contestable under data protection law. It is highly unlikely that all this data would be collected, stored, and processed in a personalized manner. A probabilistic interpretation that leaves out the individuals and their personal data and relies on a clever calculation of probabilities is much more realistic here and will ultimately be more successful.