Why does one car sell and another not? Car dealerships have obsessed over this question for decades – and they used to have pretty good answers.
If you talk to any on-the-floor car salesperson, they’ll tell you hundreds of small “tells” that reveal whether a customer likes a car or not. Maybe customers come into the store twice in one week to stare at the same car. Maybe they bring their spouse in to help them look. These signals tell salespeople whether they should raise or lower their price and how likely each customer is to buy.
When people began researching cars online, salespeople lost those tells. Most customers have already done their homework before they walk into a dealership, leaving salespeople to guess which car they plan on buying and how willing they are to haggle.
Noah John and his friend, David Pilo, created their startup Autoscores to help car dealerships understand customer behavior. After testing their theories at David’s father’s car dealership and with a handful of early adopters, they launched Autoscores in the US and Spain. Their business grew quickly until the pandemic threatened to dry up their target market altogether.
Now that the pandemic is over and car dealerships are thriving, Autoscores has ten-timesed its monthly revenue. Over 150 dealerships around the US use Autoscores to learn more about their customers and optimise pricing.
Here’s the story of how Noah and David followed the data and their industry expertise to make an app that may keep dealerships relevant for decades.
It Runs in the Family
Despite their different backgrounds, college buddies Noah and David have a lot in common. Noah was an American tennis star from Maine. David was a Spanish exchange student and the son of one of Spain’s more successful auto dealerships, Canalcar. They met as undergraduates in computer science and mathematics at Rollins College, Florida, and became friends immediately. The friendship even inspired Noah to learn Spanish.
After college, they both found jobs in tech. This was a bold move for David as his father wished him to inherit his car dealership in Madrid. David swore he would never run the family business, but he would renege on that promise later.
Instead, David got a Master’s in Operations Management from Cornell University and worked at Microsoft in the Office 360 Excel department for two years. Noah became a gameplay engineer at EA sports working on NBA Live 13 and 14.
However, feeling stuck in their careers, both friends moved on in 2013. Noah to grad school at Colorado State and David to a role as CTO and GM for his father’s dealership. David wasn’t present for the interview, but Noah shared some thoughts about why his business partner changed his mind about cars.
“I think at the end of the day it was loyalty to his family and the added freedom of being at an owner level rather than being an employee that brought him back,” says Noah.
Once David assumed the role of CTO and GM, he immediately put his tech background to use. He built a better website and introduced SEO practices to help the business rank on search engine results pages. In the process, he started thinking about how to optimize vehicle sales using his computer science background.
“The dealership had been around for four decades,” says Noah. “It had started to lag digitally. David’s dad was a ‘post-it’ guy. And quite honestly, it had served him well until the advent of digital retail. He knew nothing about YouTube and Facebook ads, SEO, PPC, or digital retailing so David took his tech knowledge and fixed that company one process at a time.”
David saw many examples where traditional thinking failed to predict customer behavior online. When people visited a dealership to view the same car repeatedly, that was usually a sign it would sell. Online, the number of views didn’t correlate in the same way.
David told Noah about these issues and they started working on a prototype to help better understand customer behavior online. They finished their first iteration in eight weeks and deployed it into the Canalcar dealership process. They then started looking for other dealerships in the US that would let them pilot their innovative but rudimentary system.
“It was a lot of hustle at the beginning,” says Noah. “I was wrapping up grad school and I remember calling tons of businesses or just walking in to talk to them. I didn’t know sales. If I’d pitched them, they’d have shut me down. Instead, I told them that I was a grad student in analytics (which was true). I’d say I have this prototype to predict car sales and I’d ask them to be a test candidate.”
Noah generated some interest initially but few dealerships took them on – even as a free trial. “It was hard to get in the door because I didn’t have those connections,” says Noah. “I remember one dealership in the Seattle area that I called. I contacted its owner in the Pacific Northwest and the guy said it was a cool product. Although he liked it, he said we should call him back when we had an actual customer. A couple of years later, I did.”
I was surprised by how receptive trade shows were to people without previous speaking credentials.
Noah says that speaking at trade shows helped win their first sales. “I was surprised by how receptive trade shows were to people without previous speaking credentials,” he says. “I spoke in a couple of educational settings and generated several inbound leads for us.”
Once dealerships began using their product, Noah and David received the feedback they needed to develop it for a wider audience. Paying customers soon followed. “I’m very indebted to the early adopters who took a raw, version-one product and made it feature-complete,” says Noah.
People Behave Strangely Online
Noah describes Autoscores as a business intelligence tool to manage inventory better and faster.
“Most traditional car dealerships will have at least 50 cars pre-owned on their lot,” says Noah. “The inventory manager or general sales manager sit at their desks every day looking at their cars and wondering how to sell them. Do they need to change the price? Do they need to market them somewhere else?”
Before the internet, Noah says selling cars was simpler. Inventory managers could easily identify the popular cars by talking to the salespeople who watched people shop for new cars. Now that people shop on smartphones before coming in to buy, salespeople have fewer insights to share with their supervisor.
“On average, people used to visit seven dealers before buying a car. Now, most go to less than two. We do what those salespeople were doing for their GMs except online. We’re championing this idea of machine-assisted inventory management – offloading work that shouldn’t be done by people to a machine so people can do the human-centric things. People can now do their job better and faster.”
Noah gave me some of the secret sauce that Autoscores provides for car dealerships: They inadvertently mimicked the airline industry’s system for pricing tickets according to supply and demand.
“Airlines look at customer behavior as they shop for tickets and adjust their price accordingly. They don’t care about historical data, they adjust their prices in real-time based purely on current customer behavior.”
But things that seem obvious tells for car popularity on the dealership floor may not be tells at all online. “Everyone knows how many times a vehicle has been viewed online. However, we’ve seen a paradox where cars with low views may sell before a car with high views. We look way below that layer to figure out why.”
Autoscores is effective on a variety of websites because it learns each dealership’s site separately.
“In our case, our AI metaphorically sits next to car shoppers on a per-dealer basis and tracks what they click and how they scroll,” says Noah. “The machine discovers different things depending on the dealer. With enough examples from shoppers, we can see which indicators mean something sells or doesn’t sell. We can give that information to dealers so they can raise or lower their prices.”
Noah shared one trick they commonly use to tell if customers are interested in buying a car:
“A click on the advance arrow of an image carousel has no significance, but if a prospect clicks back to a previous car, that’s a sign people are interested. It’s just like Tinder. Now, that doesn’t necessarily mean people are going to buy a car, but it shows a higher likelihood.”
He also gave me one misleading sign that many dealerships have trouble reading:
“Some people interact with a vehicle because they’re curious about something that has nothing to do with a car. Usually, it’s prefaced by something odd like if they immediately mess with the finance calculator on the page. That’s a sign they aren’t checking out that car. However, if they click on a car and look at the specs first, that would be an indicator they want to buy it.”
Surviving the Pandemic
By the end of 2019, Noah says they had finally added the most desirable piece to Autoscores for auto dealers: an action center that interprets the data and recommends how to respond to it.
“Analytics was a harder sell for many people. The recommendation engine was a much bigger hit. We could use data to tell a business to raise the price of one car that was popular or have a look at a car because something is happening at the point of sale. When we did that, we had a complete package and things started to take off. Then the pandemic hit.”
During COVID, dealerships faced a ton of uncertainty. Early on, federal, state, and local authorities gave mixed messages regarding whether dealerships were essential businesses or not. Noah says this immediately put their customers in survival mode.
“Many dealerships didn’t know if they were allowed to be open at all. Franchise dealerships were allowed to keep their service department running to service vehicles and so they’d keep on some sales staff too. Many people were buying cars online during this time too. But they were afraid they might be shut down at any time.”
Not wanting to lose customers, Autoscores allowed clients to suspend their accounts if they wanted – an option many took. The measure drastically cut Autoscores’ revenue but helped the business retain more customers. Fortunately, the setback only lasted for about a year, and since late 2020, Autoscores has slowly regained momentum and is now doing better than ever.
“Once dealerships got comfortable with staying open, they started thinking how to cater to customers in the new pandemic period,” says Noah. “The automotive sector was one of the earliest to adopt protocols helping customers get some peace of mind. They were some of the first to put up plexiglass at the service desks and perform services over Zoom. This may have indirectly helped us, as they became more open to change.”
What Makes Autoscores Unique
Noah thinks the success of Autoscores stems primarily from his and David’s unique backgrounds. Few programmers know about the auto industry and few in the automotive industry know about coding.
“We have a strong engineering and data-science background coupled with automotive insider knowledge. It’s hard to find all this knowledge linked to automotive,” he says.
Noah and David have no clear roadmap for the future of their business. They want to grow with customer demand – but Noah specifically requested we not use the word “agile” because he believes it’s overused.
“Most other companies are large, established, and entrenched in their processes,” he says. “Pivoting is hard for them. We can adapt to things. In the future, we’re going to focus on adapting more to inventory managers.”
Somewhat surprisingly, Noah’s biggest piece of advice for other founders is to be careful when partnering with friends or family members. “It’s seductive and easy, but it has risks. It adds a lot of stress to the relationship.”
He also suggests founders not get hung up on having a perfect product and instead favor action over inertia. Another large pitfall is assuming you know your field better than your customers.
Finally, Noah has a small list of skills he thinks every startup needs:
You usually need engineering, sales, and business strategy. You can have one person who does more than one of them but you need them all.
“You need partners that compliment your skills. You usually need engineering, sales, and business strategy. You can have one person who does more than one of them but you need them all. If you’re the tech guy you have to get comfortable with sales. Whether you’re selling to customers or potential investors you always have to sell.”
According to McKinsey¹, SaaS is most prevalent among dynamic sectors like media and retail. But Autscores proves legacy businesses are still worthy customers. Many have been running highly-profitable businesses for years. If you can help them jump to digital and prove its effectiveness, you may land loyal customers for decades.
¹The next software disruption: How vendors must adapt | McKinsey
Want to share your bootstrapping story with the world? Enter your details in this form and we’ll be in touch.