‍

We spoke with Senior Advisor Mika Laukkanen about a project where skepticism turned into success. He shares how a machine learning model became the sales team's most important tool.

Who are you and what do you do?

‍I'm Mika Laukkanen, and I work as a Senior Advisor at Data Design. In practice, I participate in various data and AI projects.

‍

You have extensive experience in utilizing machine learning and AI. Is there a particular project that has stuck with you?

That's a tough question - there have been quite a few projects over the years. I'd say using machine learning for car pricing is one of the most memorable ones.

‍

Why?

This was during a time when such solutions were just emerging, and very few initiated projects actually made it to production. This was one of those that did.

‍

What was the project about?

The goal was to leverage machine learning for pricing used cars. The idea was that it would save salespeople's time, standardize pricing, reduce pricing errors, and improve sales efficiency. As a bonus, the predicted prices could help estimate inventory value and its development over time.

‍

What kind of solution did you end up with?

We implemented a predictive model based on historical prices and available car data. The end result was an application where the salesperson enters a license plate number, and the app returns an estimated selling price along with, for example, the predicted sales time.

‍

‍

What was the reception like for the solution?

I'd say the initial reception among salespeople was mixed. For instance, they quickly spotted when the model returned clearly skewed predictions, which was actually valuable feedback for improving the model.

This also made it very clear that there's no such thing as an objectively "correct" price as the same car can be priced differently. Ultimately, this can depend on the salesperson's own experience and perspective.

Nevertheless, the application has become an everyday tool in their work.

‍

What kinds of challenges arose in the project?

What stuck with me most was how quickly car prices change. We had training data spanning a long period, but older prices were no longer comparable to current ones. We had to find and test solutions for this.

‍

Any other interesting observations?

There was clear variance in prediction accuracy between car brands. Let's put it this way: predictions worked with smaller errors for so-called premium brands. Also, newer cars got more accurate predictions than older ones, which was a fairly predictable characteristic.

‍

Could the same solution concept be applied elsewhere?

Absolutely. I believe similar concepts could work for pricing real estate and many other products. The main requirement is having enough high-quality data.

‍

What was the best part?

‍The fact that the results went into production and further development. That's when you know you've done something right.