But “once everyone started to use same technology, the advantages started to go away because then it just became, ‘We’re lowering our gross; we’re giving away profits just to turn inventory,’ ” said Jeremy Alicandri, managing director at automotive consultant Maryann Keller & Advisors.

Today, more dealers are turning to artificial intelligence to set themselves apart and boost their bottom lines.

Artificial intelligence is defined broadly and has a few distinct levels, and dealerships are still only scratching the surface of intelligent data. While not all retailers have embraced the tools, dealership use of AI can do more than enhance the customer experience; it also can bolster store profitability.

There are hurdles to clear, such as trusting a process that’s difficult to understand, but with a plethora of data, many dealers are using AI to get more efficient. Not only can they better deploy employees, whose time can be freed up through AI tools, but they also can make better pricing and inventory choices and sell vehicles faster.

Turning cars

Tom Gage, chief strategy officer at vendor LotLinx, has seen machines improve a store’s overall financial health. One potential candidate for AI assistance: the regimented markdown schedule dealerships have to move cars off their lots.

“The only lever they know is marking down the car, which has to be the most costly way to stimulate demand in the market,” Gage said.

But LotLinx, which customizes marketing campaigns to specific VINs, and companies like it can use AI to better predict when a car will age out.

Dealers should weigh which vehicles need to be turned quickly and which justify a higher price, Alicandri said. Now, dealers “actually have a computer telling them this is what you should do with this car.”

For example, if a Nissan store and a Ford store each have a used F-150, the Ford dealership would do a better job selling it, so an AI tool may suggest the Ford store price it higher.

“The tools have become so accurate now that they can actually predict how long it will take to sell, how long on the lot as well as a suggested sale price,” Alicandri said. “That’s a really strong evolution of how this transformation has occurred.”

AI’s efficiency largely relies on the amount and quality of a dealership’s data. CarStory’s AI tool, for example, can identify vehicle images, down to the trim level, without any human intervention after the initial training process, said Chad Bockius, CEO of CarStory, an inventory management app that uses predictive analytics to determine how long used vehicles will remain on the lot.

The company processes billions of photos, Bockius said. “For us to do that, it required an ungodly amount of data.”

At one dealership, CarStory found more than 300 merchandising errors in an inventory of 480 vehicles.

For AI to be functional, data is critical, said Ryan Soffa, executive vice president of business development for automotive consultant Motormindz. “The key is, do you have enough data? Do you have good data, clean data that you can actually use to build a formula to come up with something to be solved?” he said.

Historically, dealers have relied on simple averages to manage their inventory, said Bockius.

“They’ve taken all the data in the market and added it all up. Fifty percent of you are going to perform better than this number, and 50 percent of you are going to perform worse. But here’s the average,” he said. “It’s not very sophisticated. The biggest issue is, it ignores the dealer’s performance.” 
AI considers factors specific to the dealership — such as market demand, supply, competition and sales history — factors that would be overwhelming for a human to comprehend. No human should be trying to organize and analyze such immense data pools, said Len Short, CEO of LotLinx.

“It’s like, no one should be mowing their lawn with a pair of sewing scissors. It makes no sense,” Short said. “It moves too fast; there’s too much data. It’s gone beyond the ability of humans.”

Dealer Brian Benstock can attest to that.

While most questions can be answered with data in the dealership management system, it’s impossible to have staffers mine that enormous database while still interacting with customers, said Benstock, vice president of Paragon Honda and Paragon Acura in New York City. Data mining was helpful 15 years ago, he said, but today, data usage needs to be smarter and analyzed by a machine.

“The data is useless without intelligence behind it,” he said. “The intelligence enables us to serve our customers the right message at the right time.”

Still, dealer skepticism is persistent.

“Dealers are reluctant to trust it until they understand it,” said Chris Glenn, COO of Motormindz.

And it can take some work to prove AI tools out. They need experience with data to learn, and they get smarter over time. Sometimes, the first iteration of an AI product needs adjusting, Glenn said. “AI doesn’t make perfect decisions as soon as you plug it in.”

Setting expectations can solve misconceptions around AI, said Jeff Cole, managing director of engagement solutions at Digital Air Strike, an automotive marketing company offering AI tools and bots.

“It requires ongoing training,” Cole said. “Dealers will not be successful with any type of new technology if they [think] you just press go and everything is going to be perfect.”

Bockius of CarStory tells dealers to allow three months to see results. During that time, his tool gets smarter with every new piece of data it absorbs. “My performance on Day 1 is not going to be nearly as good as my performance on Day 90,” he said.

Though many dealers are reluctant to change the tried and true, others say they’d rather be on the forefront of AI technology.

“The only one who likes changes is a wet baby,” said Benstock. “We are hanging onto a world that no longer exists. The sooner we realize that, the better we’re going to be.”

View the article here.