VWE Automotive is an information provider to the automobile industry and decided to integrate data science capabilities in their operations to make sure that more meaningful data is provided to car dealerships and in a more efficient and reliable way to increase the overall efficiency of the system.
In this study we will discuss the following points which will help you in understanding the Data Science team of VWE Automotive:
- The importance of becoming a Data-Driven Organization for VWE Automotive
- Internal struggles of the company against the data science approach
- Managing expectations within any organization
- Building a working and successful data science model
- Making sure that your client will utilize the new model in their organization.
However, before we go into this detail let’s take a look at the origin of VWE Automotive and its core business to have a better understanding of the topic.
The origins of VWE
The company originally started as a physical marketplace for cars in Utrecht in 1975. Starting with paperwork (f.e. import and export documents), VWE Automotive acceded the world of data a long time ago, unconsciously of the possibilities it could give today.
However, back in 1975, there was no technology to analyze data on a large scale and since 2000 VWE Automotive has been slowly transforming to a data-driven organization. From then, VWE Automotive started to add external data sources which resulted in a huge pile of data: VWE Automotive has information about 1200 characteristics of approximately 30 million vehicles in the Netherlands.
Over the last couple of year, VWE changed into a company where data science is one of the core strategies of their organization..
What is the core business of VWE Automotive?
VWE Automotive has three 3 main propositions:
It acts as a marketplace for secondhand cars and has a physical marketplace in Beverwijk along with an online platform.
- MyVWE Platform:
This platform acts as an operating system for car dealerships and using this platform they can easily manage various tasks such as APK’s, import/export documentation, digital advertisements, inventory etc.
This is a mobile application launched by the company which allows users to easily find and view information about different cars.
VWE Automotive possesses 12,000 of in total 22,000 car dealerships in their database. Other than a car dealer, the company also has various other clients on board such as fleet owners, lease corporations, insurance companies, corporates as well as individuals.
Now we understand the core business of VWE Automotive, let’s look into the reason why Data Science is important for VWE’s business model.
The importance of becoming a Data-Driven Organization for VWE
Ewoud van Bennekom, Manager Data & Insights at VWE Automotive, says: “We are a company that controls and sells data. We put a lot of time into the quality of the data and we make useful data products out of it, e.g. reports, dashboards, data file deliveries and fully automated data inquiries. But most of the time you’re still talking about flat data”
He further added, “You want to create insights that no one else has. So when we try to help our customers (car dealers), we want to give them more than a number about how much APK’s they handled. For example, benchmarking helps customers to compare their performance to the market. Also, giving information to car dealers about the number of clients is not as useful as telling which potential clients might leave them. The goal of our data science strategy is to give our clients more valuable insights. That means that listening to our clients is just as important as creating a new data science model.
So how can Data Science be of value to VWE Automotive?
To illustrate the value of Data Science, VWE Automotive describes one of their use-cases.
Every week, VWE Automotive organizes a car market in Beverwijk where about 400 cars are sold to buyers who come to Beverwijk from Europe. The supply of cars often concerns older cars (> 10 years) with quite a few kilometers (> 225,000 km) on the odometer.
It is up to the selling car dealer to determine which cars he takes from his stock to the car market. VWE Automotive provides a service whereby the car dealer through email is advised which cars to take to the car market. This is a data science model that, based on sales of the past 2 years, can predict the probability of sales on the car market on the basis of a large number of automotive brands. The model is then applied to all active advertisements that are advertised via the VWE Automotive advertising manager. Only the cars that are most likely to be sold are advised.
By delivering the advice the knife cuts on all sides. The selling car dealer is quicker from his stock and the buying dealer can choose from an offer that better matches the wishes. As a result, more sales transactions are realized whereby VWE Automotive can take care of the entire export process.
VWE Automotive is clearly aware of the potential business value that Data Science can bring. However, applying Data Science in an organization for the first time, can bring complications.
Internal struggles of the company against the data science approach
Erwin Huijzer who is working as a Data Scientist at VWE Automotive says, “Literally and figuratively, data scientists speak a different language than the traditional IT department. The existing employees prefer to work with their usual tooling (e.g. C#) whereas data scientists rather work with Python or R. Besides, I use a different methodology and have a different process to create, test and deploy my models. Because data science is still new in our organization, there are no elaborated methods which we are ready to use. Starting data science, also means that you have to create a test environment, a security protocol and a new way of documentation.
Managing expectations within any organization
Erwin further added: “One of the major misunderstandings regarding data science is that people expect it to be a magical tool that will quickly revolutionize any company. However, this is not the case which is why it’s important to manage expectations in the organization. To manage the expectation a simple rule was used by VWE Automotive. This rule is”:
Efficiency = quality * acceptation
A key thing to realize in this regard is the fact that a model that gives information with an 85% rate accuracy might be useful for one type of business and not feasible for another type. For example, if you want to know the chances of a car selling at any particular location, the model with 85% accuracy rate is sufficient. But for a new medical treatment, the 85% accuracy rate is not enough.
Whenever any company is starting a data science project it is extremely important that they have a clear image of their expectations and what the demanded quality is.
“Another thing that we should realize is that a data-driven model needs a different approach and way of work. Data scientists tend to isolate themselves and work towards improving the working and efficiency of their model. However, the problem arises when they continue to optimize the model because at a certain point they have to meet the deadline along with results. One of the best ways to manage expectations in any organization is to make sure that different departments working on the project are exchanging feedback on regular basis.”
At last, making sure that your client will utilize the new model in their organization.
One of the main things to keep in mind if you want to make sure that the new model is being utilized is to make sure that it is being adopted. Most of the people are reluctant to change and improving your product doesn’t necessarily mean that your clients will feel the improvement too. As they get new insights because of the improved system their role as a car dealership also changes. “We ask car dealers to pick up the telephone to call a hot lead, produced by a recommender system. Most car dealers haven’t done calling before so it requires a new unusual process in their work. We try to help our customers to make that shift in their work.”
Does this sounds like an organisation you would like to know more about, please take a look at https://werkenbijvwe.nl/ or contact Ewoud van Bennekom (mailto: Ewoud.firstname.lastname@example.org) or at +31 6 20605957.
Ben je op zoek naar advies? Klik hier om in gesprek te gaan.