Every beginning is difficult, as the saying goes. From a company's point of view, everything should work immediately, but it is naive to think that data analytics and artificial intelligence can solve every problem if enough data is just thrown into an algorithm.
Initial hurdles are often immense, with ever changing data landscapes and myriad distributed systems in different departments, to no sensors and poorly maintained data. By the time a company gets to the point where it can use Big Data, it may have sunk a lot of money into finding out that it was heading in the wrong direction all along.
Also, someone must teach artificial intelligence what exactly it is that it should do. A company has to know what it wants to achieve. Projects too often start without having a goal, and when there is no goal, many projects run only to keep running. The more precisely a project goal is defined at the beginning, the higher the probability of achieving the goal.
The trend is turning to cloud technologies; companies are no longer physically building the infrastructure themselves but are renting servers and software. The advantage is time! A cloud platform can be ready for use in a few hours, and it is also scalable.
A company can start small and achieve quick success before thinking bigger.
Despite the challenges, there are numerous examples in the automotive sector where data analytics is already being used successfully.
During production, for example, it is possible to check at various points whether the bolting of the car body is in order by means of data. A technician receives feedback if a bolt connection is not optimally seated and can readjust it directly. This means that defects are not only detected during a final test, but can be repaired immediately, saving time and money.
Even without a technician, welding work can be automatically inspected elsewhere. Sensor data on welding sensors provide information on the duration and intensity of the weld so that defective welding points can be reworked. It is also possible to detect wear and tear on a machine, meaning maintenance can be carried out before a breakdown, and as we all know machine failures are a major cause of headaches in production lines. Thus, machine learning and advanced analytics can help to take control of quality problems and prevent expensive breakdown that might develop out of quality issues.
Quality Navigator – A new age of quality analytics
A completely different area is after-sales. Merging information about customers enables a fuller picture of customer requirements. Advice on possible vehicle upgrades or services can be significantly optimized here. With data analytics you can get to know your customers better: who is my customer and what can I offer him?
Other industries are more frequently confronted with “churn,” or the migration of customers by contract cancelations. This is not as easy with vehicles as it is in the telecommunications industry. A vehicle is usually purchased for a longer period and, in addition, it is unclear whether, when and what a customer actually cancels, for example, when the customer changes from an authorized garage to an independent garage.
It is in an OEM’s interest in having customers visit an authorized garage to ensure the highest possible quality in maintenance – and to remain in contact with customers. With data analytics, you can find out when a customer last visited an authorized garage and what caused the switch to an independent one. Was it because of the time factor ("My car is already four years old."), a lower maintenance price, distance to the authorized garage, or a lack of service? In each case, it would be possible to act early so that migration is prevented and brand loyalty is increased.
The potential for the application of intelligent algorithms grows with the increasing amount of data that is available in companies. The automotive industry in particular offers an unprecedented amount of data due to the emergence of Industry 4.0, the Internet of Things and connected vehicles. Automation and the application of intelligent methods make it possible to process existing data and generate added value: self-learning algorithms are already generating profit from the flood of data in the automotive industry.
A large amount of sensor technology and data collection in the automotive industry as well as strong networking with the end customer offer ideal conditions for the profitable use of predictive analytics and artificial intelligence methods. Due to the growing importance of machine learning and artificial intelligence, the automotive industry is experiencing a structural shift that requires greater investment in intelligent algorithms. It is crucial to make these investments in concrete applications that are in line with the possibilities of machine learning.