1,400+ Vehicle Estimates Daily
To stay competitive and maximize profit, auto sellers strive to find that just-right price for each vehicle. Fortunately, it’s significantly easier when selling on one of Europe’s leading auction sites. AUTOproff provides a minimum price to sellers in less than 20 minutes.
Every day, AUTOproff generates more than 1,400 car value estimates and offers sellers a minimum guaranteed price. If it sells below the minimum price, then AUTOproff covers the difference. The better the company balances the guaranteed price and the actual sale price, the more profit it retains from a sale.
Until recently, producing estimates has fallen entirely on a team of skilled vehicle professionals. But as the company has grown, scaling has increasingly become an important factor.
“Obviously there’s a huge scaling challenge with manual estimating,” said Jesper Bruun Hansen, Head of Data and AI. “Our team of estimators works seven days a week. When we wanted to grow the business, we knew we needed to do something else to ensure we maintain our promise to customers to deliver quotes within 20 minutes.”
Modeling “Automated from Beginning to End”
AUTOproff’s data science team leverages AI to speed estimating time. But with various machine learning solutions, the analytics team found it still needed to write a considerable amount of code and manually handle much of the modeling lifecycle. Most AI solutions it considered were too cumbersome for the small team to set up and manage.
In a proof of value with DataRobot AI Cloud, however, the team was able to begin generating results with little setup and no coding.
“We wanted to find a high-availability platform we could trust for modeling and performance monitoring,” said Daniel Franch, Machine Learning Product Owner.
With DataRobot AI Cloud, it was quite easy for our small team to use and quite fast from access to the first model in production.
From feature selection to model monitoring, DataRobot AI Cloud eliminates the manual parts of AUTOproff’s analytics lifecycle. Where once data scientists were skeptical about DataRobot’s AutoML, they soon discovered it cut the time to arrive at optimal models.
“To be honest, we didn’t think we’d use AutoML much because that’s our main job as data scientists,” Hansen said. “But we now use it to deploy models and it’s automated from the beginning to the end.”
DataRobot MLOps simplifies monitoring models in production, enabling the team to stay on top of accuracy. Additionally, the platform has cut experimentation time dramatically. Instead of six months, they can now release models to production in three weeks or less.
“The whole experimenting phase now resides with just one data scientist where before three data scientists were actually doing that,” Hansen said.
Essentially, the platform automates the boring aspects of data science so the team can spend more time improving data quality, understanding stakeholders’ business challenges, and supporting the demands of moving into new markets.
Shorter Time to Quote
AUTOproff uses the AI Cloud platform to create its Pricing Robot to support automated car value estimates. With a regression model, the company automates 55 to 60% of all estimates – benefiting customers, estimators, and the business.
We can provide an instant price for some cars and reduce the wait for customers. The average turnaround time for requesting a quote is shortened drastically.
With the AI Cloud platform pricing more common vehicles, the company removes some of the tedious tasks for the company’s seasoned team of estimators. Now, they can focus on the more enjoyable, rarer vehicles – and can more easily get time off.
Moreover, the collaboration between data scientists and pricing specialists removes the intuition in pricing, resulting in more accurate prices that increase the chance that estimates will come closer to sales numbers.
Price Robots for New Markets
AUTOproff’s AI-driven Pricing Robot prepares the company for expansion across Europe, especially with the recent acquisition by AutoScout24. The team has confidence in its ability to meet its promises to customers, price as accurately as possible, and trust that the platform will be there reliably.
Our pricing robots are critical as we scale to the rest of Europe. We’re able to spin up quickly when going into new markets.