(originally posted in Nov 2018)
It comes as no surprise that the auto industry is looking to replace the driver with a machine. What may be of note, however, is that drivers might not be who Michigan (and Silicon Valley) make obsolete first.
Self-driving cars are on their way. Tesla, Hyundai, Volvo and other auto manufacturers have been selling cars with “semi-autonomous” driving modes for highways for several years. Meanwhile, Apple, Google, Nvidia, and the rest of Silicon Valley are competing to build better autonomous driving systems via sensors and artificial intelligence (AI). At the 2018 Detroit Auto Show, almost every auto manufacturer has outlined plans to bring fully autonomous driving to their cars in the next 2-7 years.
Uber has made clear its intention to replace its drivers with computers, cutting their costs and reducing their liability by taking the “human factor” out of the equation. Uber has even announced a plan to purchase 24,000 of Volvo's self-driving vehicles. The writing's on the wall for Uber drivers...but does that mean they’ll be replaced first?
CHANGES IN AUTOMOTIVE SALES
According to a Gallup poll, a car salesperson is one of the least trusted professions in America, right alongside Congressman. There are simply too many stories of bad experiences like high- pressure tactics, salespeople on commission or confusing offers. Companies like TrueCar and CarMax are built on the premise that consumers don’t want to negotiate with a car salesperson. The market is ready for even further disruption.
Jeff Proctor, the General Manager at Metro Honda, said that not only is he open to the idea of computers helping his team sell, he’s actively pursuing options. In a phone interview, Proctor shared, “My goals are aligned with Honda’s. I want to ensure the best customer experience possible in order to not just sell a car, but create a customer for life. It’s absolutely my job to research new ways to better serve the customer.” Proctor said that many companies are offering more customer insights, targeted advertising options, and integrated chat systems all the time.
One of these companies, Cyrano.ai, managed to impress Proctor enough that he invested in the company and beta tested in his dealerships. “They sat down with me and analyzed transcripts of conversations from our online chat system and determined who had or hadn’t purchased a car. They correctly flagged every single sale from the chat logs I gave them.” But predicting an outcome wasn’t enough. To get his investment, Proctor said they needed to impact the sales process. So the team at Cyrano developed a system that allowed individual salespeople to better understand the personalities of customers in real time and tailor the sales pitch on the fly. “Sales went up, and probably more impressive, average time of sale went way down. That meant we had improved customer satisfaction scores. I get pitched a lot of sales tools and trainings, but I have never seen anything like this.”
Many Auto Dealerships have a chat system on their website. This allows them to initiate a conversation with a potential buyer on their site. Most of these systems are manned by a person in the dealership or outsourced to a third party company (think: a call center full of keyboards instead of phones). A growing number of these dealerships are experimenting with chatbots (where a computer interacts instead of a call center using pre-programmed responses). A given dealership could have records of tens of thousands of chats on file, maybe as many as a hundred thousand conversations from their website chat system.
What happens when those transcripts are given to machine learning experts? What happens when that chat system gets as good at talking as these other computers have gotten at chess, or Go, or poker? When that digital sales agent is as good at converting a customer as IBM’s Watson got at answering questions on Jeopardy, dealerships will flock to these new options en masse.
Science fiction has shown us humans chatting with robotic assistants for decades, so why is now different? Simply put: computers have gotten smarter. Cloud computing enables super-computer levels of computational power to any user. Natural Language Processing has developed enough that Alexa can place an order for you when just 5 years ago you threw your phone across the room trying to talk to the Interactive Voice Response (IVR) system at your bank. Lastly, AI systems have shifted from reactively analytical to proactively strategic.
In 2017 alone, Google’s Deep Mind shattered goals by winning strategy competitions like chess, Go, and a variety of video games. A single Carnegie Mellon student built an AI program that handily defeated some of the best poker players on earth. Hospitals started using AI systems to read MRI’s and diagnose patients faster and more accurately than a building full of doctors. AI is learning how to understand people, identify patterns, win competitions, and work strategically. That list is starting to look like the makings of a great salesperson.
It’s not just the computers that are ready. People are increasingly comfortable talking to machines. We ask Alexa to turn on some music and tell Siri to “call mom.” More than ever, people are engaging in text-based conversations (think: email, text messaging, facebook messenger, whats app, etc.) and prioritize efficiency of communication over deep conversation. The average consumer is spending more time talking or typing to a screen to get the answers they want.
THE RACE TO ADOPTION
Clearly more has been said about cars that drive themselves instead of sell themselves. Can companies like Cyrano.ai deliver faster than Volvo? One thing to consider is margin for error. A digital sales agent that works correctly 90% of the time is better than an average person. A self driving car that works correctly 90% of the time could cost lives. The cost of building and deploying either system also leans heavily in favor of digital sales agents over robotic chauffeurs. Finally, there is incentive. Auto manufacturers can be confident that a machine is staying on message far better than they can be with human agents. This means manufacturers can have greater controls over customer experience on a national scale while helping dealerships save money related to training, turnover, etc.
Proctor, who also drives the Honda Ridgeline in the Baja 500 and similar offroad races, is intrigued with self-driving cars, but isn’t quite ready to sit in one blindfolded. “I’m no tech expert, but common sense tells me I trust a computer to talk to me way more than I trust it to drive me around the city. If the worst case scenario happens in a conversation, I can always hang up.”
Ensuring a great customer experience is certainly an easier bar to clear than ensuring the safety of passengers in and around every car you sell. Can a company like Cyrano.ai build a better salesperson before the market moves to self driving Ubers? The race is on.