Welcome to another edition of the Car Dealership Guy Podcast Recap—a rundown of key lessons from top operators, founders, and execs shaping the future of auto retail.
Today’s guest is Eddie Stivers, President at Stivers Auto Group.
He breaks down why his own cost of people sits at 56% of gross against Carvana's 20%, and the plan to fix it within 36 months.
Eddie also unpacks how he is using Claude to catch missed fleet sales and rethink decades of gut-feel vehicle ordering.
Buy broken toys, deploy your system, and take your people with you.
Every acquisition runs through the same filter of finding an underperforming store in a good market with room to grow, then bringing the marketing package, the tech stack, the process, and, most critically, the people.
"We look for underperforming opportunities. We look for an intersection of a good market that can generally pump into a great market. We'll take our basic marketing package. We'll take our tech stack. We'll take our processes. And the most important thing, Yossi, is we always take our people."
A 1,200-person town in Arizona that was selling 12 new cars a month is now at over 100, with 776% sales efficiency.
People want to be led, and broken stores are always yearning for direction.
Walking into an underperforming store, the issue is usually that no one has given the place a path to succeed.
"People want to be led. I think that's the case in any business. And so, when you when you walk into a business that is struggling, a broken toy, they're always yearning for leadership. They're always yearning for process. They're always yearning for a way to get from here to there. And everybody fundamentally, in the car business, has lots of hope and wants to be successful. And sometimes they're just missing that path."
He said the turnaround usually just takes rolling up sleeves, getting on the desk, teaching the CRM, and showing the team at an executive level that the people running the business know how to do every job in it.
Everything starts with a new car sale, and with that, inventory is ordered for what people can afford, not what looks cool.
Stivers believes you don't have a customer, a trade, or a fixed-ops relationship until you sell a new car first. Inventory ordering flows directly from that.
"A poor sales manager is going to order what he thinks is cool. A great sales manager is going to order what they can sell at volume. And what sells in volume is not what people want, it’s what people can afford."
This, at times, results in carrying more base vehicles than competing dealers.
Six years on the Ford Dealer Council cost a lot, but it was worth it.
The time commitment to chair Ford's National Dealer Council ran 60 to 70 hours a week on top of running the stores. Pre-meetings for meetings, Sunday calls running past midnight, constant off-the-record negotiations.
“I had dark brown hair when I started and I had all my hair, but it was it was a pleasure to do it. It was honestly probably the pinnacle of my career.”
The tangible outcomes: Amending all Model E requirements, securing substantially increased floor plan assistance, and moving Ford from the back of the JD Power dealer satisfaction rankings to above average in one year.
Presented by:
1. Digital Dealer - Discover what's working in sales, service, marketing, AI, and operations. Join us September 22–23 in Detroit. CDG readers: save 25% — just use code CDG25OFF at checkout. Register today @ here.
2. CDG Circles - Connect with verified dealers, join topic and OEM-specific groups, and get real-world strategies —all inside the CDG Platform. Learn more @ here.
3. Plug - If your dealership is taking EV trade-ins or looking for used EV inventory, you need to check out @ here.
Carvana's cost of people is under 20% of gross, while most dealers are running at 46 to 56%.
Stivers said the gap between Carvana's cost structure and what most dealer groups are running is not a rounding error.
"Ernie Garcia’s cost to people is under 20% of gross. Let me say that again, under 20% of gross… I ran home immediately and I looked at my own stuff and we're like in 56%. I'm like, 'Oh god, we're dead.'"
The target is below 40% within 36 months, and he says they’re working on rewriting, rewiring, retraining, and leveraging AI across every function where a human is currently doing something a machine could do better.
AI will compress in 18 months what took 80 years to change.
The auto industry barely changed from 1917 to the late 1960s. Then, it took 40 years to get the internet at scale. Then commoditization happened over 20 years. Then remote purchasing in about three.
"AI is going to upend it all in like 18 months."
In other words, the consumer is voting for doing it their way, and dealers who don't build the infrastructure will pay for it in some way.
Lost sales data fed into Claude can tell you exactly what you should be stocking.
In a fleet meeting, the team mentioned they kept getting calls for models they didn't have in stock. The immediate question was why that wasn't being logged.
"We can put our our lost sales data into Claude and create an agent. Right now, anybody can do it. And it literally tells us what we're missing, right?"
His point: The parts department has always operated this way by logging lost sales, managing suggested inventory, and treating the shelves like a produce department where nothing should rot.
Fewer brands in more locations is the right scale strategy for the AI era.
The old model of scale was 20 brands across a metro. The emerging model, however, is three or four brands in multiple regional locations.
"I can't help but hypothesize that common shared inventory won't give us an advantage at some point. And when I say an advantage, an advantage in search, an advantage in matching the customer desire, but also an advantage because I may be able to, instead of stocking 900 Fords at a store, stock 600 Fords at a store, but on that website have 3,000, because I've got five of the stores, and I can transfer inventory back and forth."
Plus, he argues that programming and employing AI uniquely for each brand is hard, and doing it across 22 brands is harder than doing it well across four.
Scanning overnight CRM conversations at 4 a.m. is how process failures get caught before they become habits.
The general sales managers have been told to keep their ringers off at night. Their GM is awake before them, going through every open overnight conversation in the CRM across every store.
"I scan through the overnight conversations in our CRM that are still open, every single night for every single store, to catch process failures and I screenshot them and I text them to them. If they're failing in a micro-situation overnight, they're probably failing in a macro-manner across all of their digital interactions and they need to address it."
Bottom line: The micro-failures in overnight conversations are almost always symptoms of something bigger happening across every digital interaction.
The dealers who can't make the AI cost structure transition will become the next broken toys.
Consolidation in the industry has always been driven by underperformance. What's new is that the performance gap is about to be created not by sales volume but by cost structure.
"I think there is no doubt that some won't go along for the ride. I think there's also no doubt that the need for as many locations, especially in this new world of mobile service, I don't think the market is going to need 2,800 Ford stores."
Ten years ago there were 3,300 Ford dealerships. Now there are around 2,800. The self-selection that happens next (who has the desire, bandwidth, capital, and scale to make the AI transition) will determine what the upcoming number becomes.
