What can you do with the data you have gathered outside of the original function?
Sticking with the ball joint, there are some incremental functions there that are relevant to fleets, such as predictive or preventative maintenance. You can use the ball joint to sense issues with a lot of parts in the vehicle, and we have algorithms to do that. Our IP is built around taking that sensor data and saying you have a control arm issue or you have a strut issue. That becomes valuable to the automaker in multiple ways. If it's a consumer, they can send them a coupon that says come to our dealership to fix this. So, now you are driving business into the dealership and you increase customer satisfaction, which is a huge driver for automakers. And if it's a fleet vehicle, you can give the fleet manager an alert. For fleet applications we have shown it's possible to save hundreds of dollars a year per vehicle by going from scheduled maintenance to predictive maintenance.
What are some of the other use cases?
Again, in the area of the ball joint, we are developing personalization functions -- mostly for premium vehicles -- although I can't go into a lot of detail. Those functions learn how you drive and the roads that you drive on from the point of view of the suspension, and make the driving experience better for you.
ZF is synonymous with with transmissions for a lot of people. What are some of the potential uses from sensors and data from the transmission?
We have a proprietary program to help automakers that have our hybrid transmission reduce their CO2 footprint using data.
Can data be used in the context of product development?
Absolutely. We are building data sets to really characterize the usage of the product. [Automakers and suppliers] engineer products for a particular customer profile, the so-called 99 percent customer, who is the most demanding. But we don't always have the data to support that. In fact, if you look at a lot of products, they are overengineered. If we can understand what that customer actually does as opposed to what we think the person does, we can save a lot of money.
What other data services can you offer?
We have algorithms that create databases and maps of where potholes and other road surface issues are located. That data is valuable to cities because they currently have to send out crews to find the potholes. There are other competing technologies that use cameras, but they are expensive, and our [data-gathering sensors] are already installed in many vehicles.
How are you handling issues of chain of custody, hacking and privacy?
We never get between the automaker and the customer. That's sacred. We provide everything we do to the automaker and they decide how to engage the customer. It’s really up to the automaker to maintain the relationship with the customer and to get approvals and permissions. But we think it's a beneficial exchange: You basically give the data and in return you get services and a better product. I think consumers have come to expect that and will come to expect it even more.
What is the potential size of data-driven businesses?
We get asked that a lot, but it's not easy to answer. I have seen numbers from consulting firms that put the value of data from a car at $100 per vehicle a year. I would say that is not unreasonable if you count the life-cycle value of the data, including customer retention and incremental business for dealerships. If you look at the increase in retention from customer satisfaction, if you look at the service business, then it's probably not an unrealistic number, and it could probably be even more if you are talking about premium vehicles, where retention is worth so much. In the context of fleets the value could be more, without a doubt.
A lot of automakers are talking about using data to generate significant revenue in the future. What have been the main hurdles to this?
Before I joined ZF, I was chief of analytics at General Motors' Global Connected Customer Experience, or GCCX, which includes OnStar. I can't really talk about their revenue, but they have a nice number. So, I think that automakers have a tremendous, tremendous opportunity here.
Many automakers are developing their own systems. Why should they get this technology from a supplier instead?
Ours is a much more efficient model, because it's just not efficient for every automaker to develop all that IP. Why does GM need to develop a set of algorithms and Ford develop a set of algorithms and BMW, and everybody else? We can drive these efficiencies from the supply base. That’s really why we set up the data monetization function, because we realized that there is a fundamental inefficiency in how things are being done now, where every automaker sets up their own soup-to-nuts data functions that are essentially replicating each other. That's exactly the opposite of the economics of our industry.
Can you talk a little bit about ZF’s Data Venture Accelerator?
What happens in a car company is you look at how much money it costs to deliver a function such as braking, and then you try to optimize, given the financial profile of the vehicle, how much you are going to invest in that system. But the people who own the economics of the brake systems are the chief engineer and people who work on the [system or vehicle] program, but they don't really understand how to make money on the data. Then you have other people who have all sorts of ideas about what can be done with the data, but that technical capability and algorithms and owning the economics of the system don't come together. They are still two different silos. What we want to do with the data venture accelerator is fuse those two into a single team that operates like a startup. We will be able to produce technical solutions that actually make money.
How have automakers reacted to ZF's data monetization portfolio?
We have automakers that have inserted our offerings into their advanced engineering and innovation processes. That is where technologies are developed, and then vehicle programs can draw upon them for actual implementation. So, if an automaker partners with us on the ball-joint sensor data, you get a whole set of functions that the data can do. We get paid based on the functions that you decide to deploy. We will say, If you want this function, this how much money you get; this how much money we should get. It's a very productive and transparent conversation.
What areas of data monetization are most promising in terms of generating revenue?
We have three focus areas for data monetization. The first is field data monetization, in which we take data from the vehicle, add value and give it to the automaker. The second is reducing verification cost for ADAS (advanced driver assistance systems). The cost of validating the systems is huge because you need to road test them more and more and put them into more edge cases. We have developed technology to collect data on one program and then reuse it in another. A third focus area for us is CO2 reduction. ADAS verification is probably the easiest to get traction with an automaker quickly, because it’s a known pain, and it has got a number associated with it. That is followed by CO2 reduction, another known pain. The potential is great for field data monetization but in the auto industry, people always would rather save a dollar than make a dollar. And that's in the "make a dollar" category; the first two are in the "save a dollar" category.
Who is your competition in data monetization?
The competition is the really the traditional mindset of the automakers that basically says, We need to do everything ourselves. I think what we are realizing today, like with the examples of brakes or ball joints, is the economics work a little differently and that there is a great efficiency in partnering with suppliers that understand the systems really well. When you think about it, data is not gold, it's just another asset in your portfolio. You just need to approach it in an economically effective way. Automotive companies used to make seats, and they even used to make the springs in the seats, but they realized that it's probably cheaper if somebody else does it because they have scale. It’s the same with an algorithm to predict brake wear. We don't need 10 automakers developing similar algorithms. That just delays our industry and delays delivering value to the customer.