Can a car company know more about your next Mercedes-Benz purchase than you do? Daimler AG, which makes the iconic vehicle, uses a combination of artificial intelligence (AI) and predictive analytics to accurately determine when their existing customers are ready for another car.
Daimler's Mercedes brand wanted to get more value from its data when it came to first and second car purchases. The automaker was interested in what factors helped a customer choose an entry-level A-Class vehicle as their first purchase, as well as what factors might cause an existing Mercedes luxury S-Class owner to purchase an A-Class vehicle.
When we first decided to innovate with analytics, we set our sights on predicting future A-class, entry-level car buyers," says Daniel Pape, a senior project manager at Daimler AG. "We knew that in order to gain real value from our data, we needed to leverage artificial intelligence (AI)."
The company prepared its project by looking at a combination of customer relationship management data culled from retailers and campaign management.
Daimler started its data modeling in Berlin with a heat map to predict in which area the highest potential A-Class driver would buy a car. Daimler had 100,000 customer records in its database. An initial round of analysis whittled down that number to 5,000 potential buyers of an A-Class Mercedes. In six weeks, they found dozens of potential customers who were most likely to purchase an A-Class vehicle.
The team dug so deep into the data and investigated these customers so thoroughly from this dataset that they identified a handful of customers who owned two S-Class Mercedes vehicles who were considering purchasing a third.
"We set out on a path of trial and error with many algorithms and many predictions," says Pape. "Most importantly, we needed to find a way to uncover whether our algorithms and predictions were correct—and if they were giving us correct results."
In the video below, Pape discusses how Daimler uses AI and predictive analytics to predict customer car purchases.
One challenge that the Daimler team experienced was being uncertain that their algorithms were sometimes faulty. By continuously applying human intelligence to the process, Daimler found its predictions getting better and better—and ultimately this helped the company discover thousands of future car buyers.
"We had a lot of algorithms, but we had to figure out if it was accurate or if the information was of true value since sometimes customer queries are not complete or truthful," says Marc Winter, a consultant application management at Daimler. "We began by deep-mining the data, and then utilized business intelligence (BI) and AI. Using the functionalities, we returned to BI to challenge our models. With an open solution, we were easily able to visualize our results."
Winter noted that Daimler could pinpoint whether one area or the algorithm needed to be tweaked. By visualizing our predictions, we could answer critical questions like: What are we predicting? What amount are we predicting? Why are we making these predictions?
Daimler used a combination of AI and analytics to test out their algorithms and visualizations in order to validate their predictions along with deep mining in their datasets. The combination allowed Daimler to come up with a composite sketch of a customer who was an S-Class owner who might purchase an A-Class Mercedes.
"It was really interesting to challenge a machine with human interests and human understanding of customers," Winter added.
The next big thing for Daimler is to apply continuous intelligence to its processes. This means the company will get a feedback loop of the artificial intelligence and how it relates to the customer decisions so that the algorithm can learn from itself.
In the video below, Daimler's Marc Winter discusses how the company uses AI, deep mining, and continuous intelligence to gain insight and confirm its conclusions.
To learn how you can benefit from Oracle Analytics, visit