Markus Sontheimer has been DB Schenker’s Chief Information Officer (CIO) since December 2015. In June 2016 he was appointed to the additional role of Chief Digital Officer (CDO). He has since consistently advanced DB Schenker’s digitalization strategy. He is convinced that “In three to five years we’ll clearly be perceived as the trailblazer of digital transformation.” Photos: Michael Neuhaus
Mr Sontheimer, there is frequent hand-wringing about volatility in the airfreight industry – but surely it is not the only sector having to deal with this problem?
It is true that for airfreight operations, capacity bottlenecks are especially critical for the customers affected, due to immense time constraints. More generally speaking, though, volatility does play a key role with every type of freight product. It is something that keeps the industry, as well as my own team, on our toes all the time. To better understand volatility, it is useful to study the various industry sectors. It then becomes apparent that fluctuations in demand are often cyclical in nature. Take the automotive sector, for example: if sales for OEMs in Europe take a hit, or conversely, if there is a significant increase, then this will be felt very quickly – even in the warehouses. Other instances have more to do with seasonal fluctuations. Think of the electronics industry, with the enormous peak they experience in the run-up to Christmas. For us as a logistics service provider, volatility is sometimes also felt in the sense that the volumes originally requested by the customer differ vastly from the volumes that are actually purchased in the end.
You said volatility was also a constant preoccupation for your own team. What are the digitally-based “recipes” that you might have at your disposal to deal with it?
Given that it can be difficult for many customers to predict their own future requirements, we see it as our job to do everything in our power to arrange for the necessary capacities to be available. For some time now, we have been tackling this issue much more systematically than we did in the past, and we do this mainly in the context of predictive analytics. Using historical data, we develop and train algorithms that allow us to project volume trends. These forecasts are constantly compared with the actual values, and doing so helps to optimize the algorithms even further. The goal is to bring the probability of making correct forecasts up near the 95-percent mark. As a matter of fact, we are already using solutions for land transport and ocean freight, and they are proving very successful.
Teamwork! At its headquarters in the German city of Essen the logistics service provider employs numerous specialists who develop data-driven solutions that, among other benefits, significantly help reduce the effects of volatility. Photo: Michael Neuhaus
Please give us an example of such a land transport solution.
Well, one of them allows us to predict the volumes that will be arriving at a land transport terminal some four weeks in advance. The underlying algorithm is powerful, and the accuracy of the forecast is high: for a period of 20 days in advance it already reaches the 95-percent level I just mentioned. We provide the figures to the branch managers, so that they can adjust their capacity planning accordingly. The figures are based on our own historical data, which in this case is adequate. In other instances we also make use of external material such as traffic or even weather data.
How good is this tool when it comes to ocean freight?
Ocean freight tends to be very volatile, and that impacts on freight rates, which are subject to correspondingly large fluctuations. But customers prefer to work with fixed prices for specific periods of time, like a whole year. This is why we have developed an algorithm that helps our trade lane managers to work out pricing structures. Not only is this what customers ask for; it also helps us to operate efficiently – a win situation for both sides. Incidentally, we use this same predictive pricing for land-based transportation.
Are there similar tools available for contract logistics?
This is where we make use of so-called location-based analytics – in this case based on data derived from WIFI and video recordings. What we look at in particular here are the movements that happen within the facility, and we can then optimize the warehouse design based on this information. While this does not relate directly to volatility of the kind that generates fluctuations in capacity utilization, it does help us absorb peaks in the volume of incoming goods.
Connected! Born in 1968, Markus Sontheimer has a bachelor’s degree in Industrial Engineering. Prior to joining DB Schenker, he held several top positions at Daimler AG and at Deutsche Bank. He has a daughter and two sons, collects glasses and enjoys playing drums. Photos: Michael Neuhaus
Who develops these solutions?
We do that ourselves! We hire specialists for this purpose, for example here in our headquarters in Essen: mathematicians, physicists as well as computer scientists. They tap and analyze our so-called data lake, and then build the algorithms on which the services are based. To achieve maximum benefit, they cooperate closely with the business units. In the course of our development projects, we were able to determine quite clearly that the algorithms were always better than the gut feeling of our managers. Yet an important point I want to make here is that we regard these solutions as decision support tools for our managerial staff: their know-how is and will remain a vital factor nevertheless.
Is it conceivable that predictive analytics tools using Big Data could be developed that forwarders could use together with shippers, and even carriers?
From my point of view, the answer is yes. The only question is: how can you make sure you cover the entire market? And I mean both in terms of the demand for volume and the availability of capacity. Because these two areas are subject to influences stemming from market participants who are completely unrelated to each other. Manufacturing companies, for example, will sometimes make decisions about their production programs and about changes to the supply chain at very short notice. And carriers may decide to make capacity adjustments on certain trade lanes at equally short notice. For these companies, this kind of data is highly sensitive and therefore not publicly accessible or transparent.