The relationship between OEMs and suppliers is undergoing vast changes, that much is clear. One person that is particularly aware of this is VDA professor Robert Dust. In this interview, he tells us how this transformation is occurring and how Google and Apple are involved in it. He also answers our core question: What does quality really mean?
Professor Dust, you’ve held this chair endowed by the VDA since 2014. Your department at the Technical University of Berlin is called “Quality Strategy and Quality Competence.” But: What does quality really mean?
The term quality, as defined in standards, is certainly outdated. It comes from a time when the discussion revolved around concreteness and tangible objects. However, in the future, this term will be translocated into the world of data. There will be no more traditional target/actual value comparison. Earlier on, the target value was set by producing companies, but in the future it will emanate from various networked sectors and be more customeroriented. There will be completely new partner networks, and car manufacturers will no longer have complete power over their own products. Take Google, for example: You cannot give a specification sheet to a Google supplier. This is because, as the manufacturer, you no longer have any influence over how Google will function in your cars and what quality criteria it will comply with. In short, the term “quality” will move away from the purely physical, corporate meaning and take on more customer-oriented significance. Asking what customers want is more important than ever because customers are now becoming more active at a faster pace, and they refuse to be told what quality means: They define it themselves.
What will become of traditional suppliers that simply want to sell their parts?
They won’t die out immediately. I am certain that two types of supplier will continue to exist: the traditional supplier who supplies components and the type of supplier who earns money with new products and services, such as software. This in turn poses the question of what the business models of the future will look like. It will be analogous to consumer electronics – these days, nobody makes money with laptops, monitors and keyboards anymore. Software and intelligent data handling are where the big money is.
How does this change the relationship between customer and supplier?
There will still be powerful car manufacturers, exactly as there are today. Equally, the suppliers seeking business here can still make money with products for the big OEMs alone. However, suppliers are consolidating more than ever and therefore have more power. For example, an OEM has to communicate in a completely different manner with a systems supplier than they would with a medium-sized screw manufacturer. The same applies in reverse. A chemical company which delivers one ton of granulate once a year is definitely less liable to allow a car manufacturer to put them under pressure.
Total Supplier Management is one of your specialist subjects. What should we know about it?
We know that suppliers can create external added value at a rate of up to 90%. Therefore, if a car manufacturer receives poor quality supplies, then not much can be achieved as they are not producing anything themselves but rather integrating various components. So this raises the question: How can I control my supplier network? If we look at how companies are managing their supplier chains today, then tools such as Excel pop up straight away. However they can only be used to look back at what has been done. This is not risk management as I define it. The preventive and integrated approaches are missing. If you want to practice Total Supplier Management, then you need to monitor all suppliers and not just the renowned top 50.
What needs to change?
We can work with trends and prognostics right now. The models for this already exist, you just have to apply them. This is where suppliers of intangible products come into play again. Google or Apple software is not delivered to a warehouse: You have to take what you’re given. You only experience it for the first time when it has already been integrated into the car: The customer experiences it first. Therefore, the supplier is no longer a supplier but becomes a partner on a level playing field that you have to cooperate with.
Partners in this sense are new: They never existed before.
That’s right, and there are other examples, too: Take the charging infrastructure for new electric vehicles. Who will install the charging points at the highway rest stops? The car manufacturer? The charging points are not integrated into the car and yet we cannot do without them. Or think about content providers, such as map and navigation services. Someone has to take care of them as well. And the automotive industry would do well to get involved here before other entities do. Even if this means extending beyond the traditional business sectors.
What is still going wrong in supplier management?
Today, most companies collect significantly more data than they can actually analyze. We don’t need tools that just produce pie charts, we need tools that generate real knowledge. Many tools are lacking intelligent data analysis, even though there are already algorithms for this. In addition, very few companies ask this question: Why are we doing this?
Risk management isn’t about coding data into a traffic light system, it’s about asking what happens when the light is red. What action should they take then? Many companies just measure risk and are content that it’s been measured. This, however, just leads to a typically German question: Who is to blame? Our studies prove that it’s not always the supplier’s fault. This means that data is there but we’re not using it correctly? There’s already more than enough data today, but of course everything comes down to data quality. In our department we have developed a data checker. We have used it to prove that around half of all data collected is not valid. This does not necessarily mean that the data are wrong but that they simply are not useful.
Can you give us an example?
Many companies measure their number of complaints. However, this number is not useful for trend or prognostic calculations because it is not a basic key figure. If I have ten complaints in January and five in February, we might think that we have improved. However, if I only received five parts in February in total, then I have a failure rate of 100%. If 1,000 parts were delivered in January, then the failure rate is 1%. This shows that a key figure like this is not appropriate for predicting trends and prognoses. Therefore, a basic requirement is that data need to be whipped into shape before processing. There is a lot of data that is useful but for much data you have to pose the question: Why have I got this data? What is its purpose?
So everything you need is already at hand?
Of course there is also a lot of data that are still missing. There are also meaningful key figures for which algorithms have not yet been created. There are examples for all of this: too much data, too little data or data that is missing altogether. The most important thing is to establish this so that the relevant analysis can be carried out. Intelligent algorithms can only be applied to data once the data quality is at an appropriate level. A valid database is the basic condition you need to establish for actually implementing a productive IT solution.
It sounds like there are excellent prospects for the software sector. I’d like to thank you for agreeing to this interview, Professor Dust.
It was my pleasure.