whynotroman

News · Oct 16, 2026 · 12 min

What auto and AI leaders posted, October 16. Google reviews decide which dealer AI recommends

Posts from the last day on X and YouTube. Google reviews shape which store AI recommends, prime buyers shop by payment too, Tesla shows the new Roadster, and VW, Stellantis and Renault team up against Chinese brands. My take on each post.

Every month more buyers ask AI, not a salesperson. Which store can I trust, what will I pay each month, where is the car near me. Today leaders talk about where AI gets its answers and how a dealer gets into them. And last night Tesla showed the new Roadster, so half of the feed is about it.

News from the trade press is in the morning issue. AI tips are in a separate post. I added my take to each post.

What dealers say

Google reviews teach AI which store to recommend

Brian Kramer of Cars Commerce spoke on the Car Dealership Guy show. He says AI uses Google reviews as one of the signals when it picks a store for a buyer. Good customer experiences give AI more evidence that a store can be trusted. "All dealers are training the LLMs as to how the LLMs should treat you moving forward."

My take. A bad review used to reach ten people. Now AI reads it and repeats it to everyone who asks "where do I buy a car in my city". A same-day reply to a review is now part of sales, not a monthly task for marketing.

What you can do today. Open your store's reviews from the last month. Reply to every review that has no reply. In a reply to a bad review, give the name and direct phone number of the person who will fix the problem.

Car Dealership Guy post on X

109 Carvana listings on Google point to other people's addresses

CDG News looked at Carvana listings on Google. There are about 266, and 109 of them point to addresses that are not Carvana sites. These are government buildings and other places in 36 states.

Those listings have 630 reviews, about two thirds negative. Shoppers who drove to those addresses wrote part of them. Carvana says the listings help local shoppers find cars for delivery or pickup in their area. Nicole Santana of PureCars says the pages can lift Carvana in local search and in Google's AI answers. This works even where it has no store, and it pulls buyers away from local dealers.

My take. Being close no longer protects a dealer in local search. Someone can take your area with one pin on a map.

There is one answer. Your Google listing must be the most complete one in your area. Photos, hours, cars in stock, fresh replies to reviews.

Car Dealership Guy post on X

Prime buyers shop by monthly payment too

Per FlexPath DXP, 78% of prime and super-prime shoppers shopped by monthly budget. For subprime shoppers it is 80%, almost the same.

Prime shoppers were also 1.5 times as likely to prequalify for a specific car, 44% vs. 30%. And 81% of those prequalifications started on search results or car pages, versus 7.5% on finance pages. FlexPath DXP CEO Tarry Shebesta says it like this. "They're asking what a specific vehicle will cost them each month."

My take. Buyers think in payments, even with good money and good credit. If the payment hides on a finance page, nobody sees it. Put the payment next to the price on every car page.

Car Dealership Guy post on X

AI writes the draft, people check and send

Ben Sporn, CMO at VIP Automotive Group of Long Island, says AI helped one e-commerce team triple its customer replies without cutting jobs. AI drafts a reply from the order data, and a team member checks it and sends it.

My take. This is the calmest way to start. Salespeople are not afraid of losing their jobs, and buyers get faster answers. After a month you see which replies people almost never change. Those you can give to AI fully.

Car Dealership Guy post on X

Dealers who spend 90% on search show up too late

Car Dealership Guy released a talk with Jason Goldberg, who leads automotive at The Trade Desk. You can see his main points in the chapter list.

71% of brands are chosen before the buyer searches. Move budget away from 90% on search. A 60/40 split between brand and performance gives three times the results. A dealer who waits only for the moment of purchase becomes one more seller in the crowd.

My take. When a buyer types a model and a city into search, they have almost decided. You pay for that click with five other dealers and then compete on discount. The winner is the dealer the buyer knew before the search.

Car Dealership Guy post on X · Episode on YouTube

Women in dealerships like the career but not the pay

In a survey that Car Dealership Guy shares, 82% of women in dealership roles say they are satisfied with their career. But only 56% feel their income matches their effort. For men, that number is 73%. To support women, 42% pointed to leadership training, 39% to continuing education, and another 39% to hiring more women.

My take. A 17-point gap is a question about the pay plan, not about women. If the plan gives the good customers to a few favorites and the rest to everyone else, people leave. A clear way to share leads and a clear pay plan keep people better than any training.

Car Dealership Guy post on X

What carmakers and the market say

Tesla shows the new Roadster with 0–60 mph under one second

Last night Tesla showed the new Roadster. Sawyer Merritt lists the 0–60 mph times. The Performance trim is under 1.6 seconds, Thrust under 0.9 and Rocket under 0.59 seconds. Tesla says the new battery and drive unit give 1 MW of power. That is one third more than the Model S Plaid, with half the mass.

Elon Musk wrote "Turns out the future does have flying cars". Car Dealership Guy called the show a "home-run".

My take. Few people will buy a Roadster. But Tesla again showed how to sell with an event. Millions watched a car they will not buy and remembered the brand. A dealer needs its own reason for people to visit the store without a plan to buy today.

Sawyer Merritt post on X · Elon Musk post on X

VW, Stellantis and Renault team up against Chinese brands

Automotive News reports that Volkswagen Group, Stellantis and Renault Group are forming their own "Big 3", like Detroit. They will ask the EU for local-content rules and relief from EV mandates. Chinese brands doubled their share in Europe to about 10% from January through August.

VW CEO Oliver Blume put it like this. "We sell large numbers of cars at relatively low margins. This business model is under enormous pressure in Europe."

My take. When three giants go to the regulator for protection, they are not sure they can win the buyer in a fair fight. For dealers in Europe this is a signal. A Chinese brand in the store next door is here to stay. Prepare your salespeople and your trade-in prices for it.

Automotive News post on X

Stellantis will build Jeeps in China and sell them worldwide, but not in the U.S.

Stellantis will build Jeeps in China again from next year, after a four-year break. With Dongfeng it will make two electrified Jeep models and two Peugeot models. The tech will come from Dongfeng, and driver assistance from the Chinese company Momenta. The cars will go to China and to world markets including Canada and Mexico, but not to the U.S.

My take. An American brand on a Chinese platform shows where EVs are cheaper and faster to make today. A Jeep buyer may soon get a car that is American only by its badge. Salespeople need to explain honestly what is inside and why it is not a minus.

Automotive News post on X

Ford collects interest in a sub-$30,000 pickup a year early

Car Dealership Guy reports that Ford opened early registration for the Fathom electric pickup. Official pre-orders open in early 2027. For now, buyers leave a name, email and ZIP code on the model page. The price will be under $30,000, while the average new EV costs $56,238, per Cox Automotive.

My take. The carmaker builds a list of buyers a year before the car. A dealer can do the same. Keep a list of interested people in your city and call each one when price and dates are known. Dealers who start now will get the first deals without ads.

Car Dealership Guy post on X

Carmakers feel better about the future, dealers do not

The Automotive News Auto Industry Confidence Index for Q3 fell 1.4 points overall. The reasons are tariffs, rising costs and program delays. Carmakers feel better, with their score up 2.4 points. Suppliers and retailers feel less positive.

My take. The carmaker sees the plan, and the dealer sees the lot and the buyer's payment. When their moods split, the dealer gets a sales target that does not match the market. Have your own forecast for your city before you talk to the carmaker, not after.

Automotive News post on X

AI and sales

Lemkin says to put your first agent on the contact form

SaaStr founder Jason Lemkin names the simplest place with the best return for an AI agent. It is the "Contact me" form on your website. "No one should wait days to hear back from a 21 year old human to qualify them."

My take. In a dealership the web form is often the slowest place. A lead comes in at night, the salesperson sees it in the morning, and the buyer is already at another store. An agent replies in a minute, asks two questions and books a test drive. The salesperson only has to meet the buyer.

What you can do today. Send a lead to your own website at 9 p.m. and time the reply. If it takes more than five minutes, start your AI work here.

Jason Lemkin post on X

For people who work with AI

Mollick says to ask AI for interim results and tests

Ethan Mollick writes that the model's chain of thought, which you could read before, now explains little. An agent works long and alone. To steer it and see what happens, ask for interim outputs and tests, and use AI to help fix the problems you find.

My take. It is like a new salesperson. You do not wait for the end of the month, you check each deal step by step. Do the same with an agent. Ask it to show the result after each step, not only at the end.

What you can do today. In your next task for AI, add one line. "After each step, show me what you got and check it before you go on."

Ethan Mollick post on X

Mollick says to put the best model in charge of cheaper ones

Mollick points to a paper. Small gains in accuracy and judgment add up over long tasks and create huge gains in results. So he doubts that frontier models are easy to replace with cheaper ones. He suggests using the best models as orchestrators.

My take. Saving money on the model that talks to your buyers is a bad idea. One mistake in price or delivery date costs more than a year of subscription. Give the rough work to a cheap model, and keep the main task with the strong one.

Ethan Mollick post on X

Levie expects background agents to use 100 to 1,000 times more tokens

Box CEO Aaron Levie writes about the move from chatting with an agent to agents that work in the background. He thinks token use will grow 100 or 1,000 times. Today work is limited by how much information a person can process. Soon you will give a task to ten agents and say "come back when you've accomplished this goal". He says most of this work will be new, not a replacement of old work.

My take. For a dealership this means checking every deal, every call and every lead, not a few once a week. Today a manager listens to five calls out of a hundred. Soon an agent will listen to all hundred and bring the five you need to review.

Aaron Levie post on X

In brief

  • Electrek says EV sales in Europe soar, while North America drops 44%.
  • Electrek reports that Slate opened its first test drive center in Chicago, in a former Tesla showroom.
  • Electrek writes that Tesla remote operators keep crashing its robotaxis.
  • OpenAI Developers says Sites, a tool to build team apps, is now available for all Enterprise customers.

Also on video

Next note · 11 minWhat auto and AI leaders posted, October 15. Mixed prices on a dealer site confuse AI →Posts from the last day on X and YouTube. AI tells buyers to call for price when dealer data does not match, 73.9% of auto loans get approved but deals get riskier, and an AI agent cut SaaStr's late invoices from 56% to 8%. My take on each post, with a link.

Roman Spitsyn. I build sales teams where AI agents do the work and people close the deals.

Read next

← All notes