The Return of the One-Rep Deal

AI agents are taking on more of the repetitive work involved in dealership lead response and follow-up. Jim Lawrence considers whether that could allow one salesperson to manage a customer relationship from first contact to vehicle delivery, with more time for the human interactions that matter.
The Sales Floors’ Revolving Door Is the Profit Leak in the Business Model
Every dealer knows the number, and most have stopped being shocked by it. The National Automobile Dealers Association (NADA) Dealership Workforce Study reported 66 percent annualized turnover for sales consultants and 42 percent across all dealership positions in 2024. NADA has since released a 2026 study using 2025 data. So, in short, dealership staff turnover still sucks.
Then there is the productivity side of the ledger. The "10-12 cars per salesperson per month" trope is regularly attributed to NADA, as in this CBT News piece on the Mission 22 program. An older NADA-sourced benchmark shows it has barely budged in a generation of technology iterations across CRMs, DMSs, and BDCs. We spent “Billions and Billions…,” - Carl Sagan, to little effect. AI is considered the next tech song on the productivity dance card. Hands on hips, now begin the box step without stepping on your partner’s feet.
Those two numbers are connected. We hire people, sort of train them, then ask them to chase hundreds of internet leads with fake phone numbers, bad email templates with stale verbiage and crappy imagery that end up in spam folders because of their filthy dirty HTML, hand them a floor schedule and a draw, and then act surprised when they leave, sometimes after they have sold vehicles to their friends and family. The grunt work is not an unfortunate side effect of the job. It is a major reason the job burns people out.
What Dealers Running Agentic AI Actually Report
Glenn Pasch's new Pasch Group research, AI and Dealership Staff: Finding the Right Balance, commissioned by Impel, (So grain of salt time) offers directionally useful field data on this perpetual issue plaguing dealer staff management. Author's Note: HOWEVER, it is qualitative, built from interviews among a likely managed sample of Impel customers between June 3 and August 25, 2026, and the report itself says it should not be treated as statistically representative of the industry. So, I recommend reading it as well-documented dealer experiences, as anecdotes with a theme to share. Like me!
Within those limits, the model is reliable:
- The Sales Support AI Agent owns the text and email grind; humans own the phone and the personalized video. Dealers who held that line got the best results, and those who let staff jump into the AI thread reported customer confusion and declining engagement.
- No dealer in the sample fired anyone because of the productive friendly neighborhood AI Agent. Half reported smaller headcount through ordinary attrition (which we know is huge relative to other industries). One GM took per BDR lead management from 100–125 a month to 240–250 with no decline in performance metrics. Another saw declining number of BDC agents earned 20–25 percent more because the same appointment pool was split among fewer people. An Illinois group trimmed a service BDC from 31 agents to 15 through reorganization rather than termination. Author’s Note: I wonder what the attrition rate is for the “reorganized employees?”
- After-hours consumer support is where Agentic AI tends to shine. The report cites data that 40 to 50 percent of internet leads arrive outside business hours, and 90 percent of the dealers interviewed named after-hours capability as a primary motivation or the clearest defensible value.
- Some things stay human, without exception: financing and F&I conversations, complaints and service recovery, returning buyers, negotiations, and any customer who asks for a person.
Notice what is missing from that list: nothing says a salesperson should be a message-chasing monkey. The research describes a world where humans show up when there is something valuable that they can contribute to the process and shared by their helpful AI Sales Assistants, R2D2, C3PO. As one Midwest GM put it, "Let AI beat you to the customer." Then the salesperson picks up where the conversation left off.
The Thesis: Fewer Reps, One Relationship POC, Start to Finish
Here is where I will take one step beyond what the research supports and make a prediction. If agentic AI absorbs lead response, long-tail follow-up, scheduling and data hygiene, the work that remains for a salesperson is the work that earns trust: the call, the video walkaround, the handoff that references what the customer already said, the deal itself. Those are the touchpoints of value for H2H engagement. Further, with a bit-o-Auto Retail Compliance...and a little dash of Truth-in-Advertising, this new “old-way” of H2H selling, Price Discovery Is likely Inbound. Ready or not, something transparent this way comes.
When those are the only touchpoints a sales rep relays, the BDC-to-Internet-to-floor relay race starts to look unnecessary. One well-trained person can own the customer relationship from first lead response to vehicle delivery, supported by an AI team that never sleeps, never forgets Day #42 follow-up, the "Answer to the Ultimate Question of Life, the Universe, and Everything" associated with automotive deals and can help your team send compliant email templates at 4:45pm on a busy Friday evening, with appropriate human review.
A single POC at the dealership is not a new idea, by the way. As I suggest above, it is how car selling worked before we layered-on specialized departments to manage the volume of activity that people could not handle as the modern tech environment evolved in automotive retailers operations. Much of that repetitive administrative work will fall into the laps of the helpful AI Sales Assistants, leveraging the staff time available to speed up the deal close.
Illustrative math, not a forecast: a store with ten reps averaging 11 cars a month sells 110. The same store with six AI-supported reps would need to average about 18 to match it. That is a stretch against a 10–12 average, but a long way from the 22-a-month aspiration, and it comes with lower recruiting costs, fewer onboarding cycles and fewer people leaving after six months. Even if productivity gains land short of that, a smaller, better-paid, more tenured team is a better bet than the current cost of churn from the revolving door.
Two caveats to be aware of. First, the research is clear that the handoff is where value is won or lost, just as in today’s current sales process. The staff who choose to skip the AI2H conversation force customers to repeat themselves, yet another time-sink and enraging feature of current sales process design. A key sore spot that consistently ends up in CSI scores. A one-rep model only works if the rep reads the file. With apologies to Yoda: If this requirement their compensation package reflects, the file read they will.
Second, compensation matters. Compensation ALWAYS MATTERS. Dealers in the study who told staff in writing before launch that they would receive full commission credit for AI-set appointments saw less resistance than those dealers who delayed committing to such a commission policy. We’ll see how long this commission deference lasts, but how sweet it is, for now.
Ready or Not, The Customer's AI Agent Is Already Walking Your Virtual Showroom
The other half of this picture is on the buyer's side. Cox Automotive's research, as cited in the Pasch report, found that 25 percent of new-vehicle buyers engaged AI tools during shopping in 2025, and those buyers reported 84 percent satisfaction with the buying process, among the highest in the study. Buyers who can research, compare and shortlist with an agent will skip steps that dealers have spent billions over decades engineering.
Meta's new Muse agent provides a good preview of what comes next, and of its limits. Muse launched on September 8. On September 20, Amazon blocked it from shopping on Amazon.com, saying Meta had not given notice and that the agent does not identify itself while browsing; hmmmm… I wonder why? Meta says Muse runs in a secure virtual machine and cannot see passwords or payment details. The dispute is still unfolding, and Meta has since announced additional retail partners, including Walmart and Best Buy. The lesson for automotive is not that agents will fail. It is that the major platforms will decide which agents get into which walled and proprietary technology gardens, and a dealer's customer may arrive via one agent that is welcome and another that is not.
And customers have reasons to want help. Gallup has long put car salespeople at the bottom of its honesty and ethics ratings. In Gallup's 2025 Honesty and Ethics poll, 7 percent of respondents rated car salespeople's honesty and ethical standards high or very high. Telemarketers were at 5 percent and members of Congress at 7 percent. Lobbyists were not surveyed in 2025; their 2024 result was 4 percent. The professions ranked alongside car salespeople:
- Telemarketers
- Lobbyists
- Members of Congress
So, the "bad apples" running dealerships so vilified in automotive publications LinkedIn posts over the years have won. If you are ranked with the Congress-folks and the Lobbyists who own them, the game is up. And it has only gotten worse since the $10k over MSRP days of covid usury, aka daylight robbery.
Negotiation is where that distrust lives, and a buyer's agent is a very sensible way to avoid it, and the traps dealerships have built over decades to corral you on the lot and roll you over.
I Coined “Personal Purchase Agents,” Now Meet the Protocol Droid Problem
Think of a Personal Purchase Agent (PPA) as C-3PO: fluent in six million forms of communication, excellent at relaying information, and not the one you send to negotiate with Jabba the Hutt. That usually requires a Thermal Detonator. A PPA can search inventory, compare out-the-door prices (California’s CARS Act has its own requirements; the FTC withdrew its separate federal CARS Rule, while other federal consumer-protection laws still apply) and even haggle over numbers. What it cannot do is earn the buyer's trust in a store, explain why this trade appraisal is fair, or catch the anxious look when the Finance & Insurance conversation starts. Those are H2H moments, and the Pasch research found dealers keep them human for exactly that reason.
There is also a trust trap in the AI-assisted process itself. The report describes one manager who rolled out AI without telling customers, and who later heard about customers feeling "bamboozled" when they discovered who they had been chatting with. If our entire value proposition is H2H trust, disclosure is not optional. It is effectively the 4th law of robotics, i.e. announcing I am a droid. I hope you are cool with that.
What Should Dealers Do Now
- Write down the problem you are solving before buying any AI Agent platform. The report's most consistent finding is "problem first, technology second." Cox data cited shows 69 percent of dealers expected AI to drive sales growth, but only 22 percent reported seeing it.
- Redesign workflows before launch, not after. One GM in a large public group saw a 19 percent contact-rate improvement after redesigning around AI. Sister stores that did not get 3 percent.
- Define the H2H moments. Map exactly where a human steps in, who it is, and how fast. Top performers in the study moved in three to 15 minutes after a handoff alert.
- Name a Champion. One accountable person for the knowledge bank, conversation audits and staff communication.
- Be agent ready. As I wrote in Got Agent?, structured, accurate, dealer-owned data is what lets a buyer's agent find you, your website and quote you correctly back to their human overlords.
- Rebuild the comp plan and the job description around the human work. That is, the essential H2H parts of the sales job worth keeping, avoiding the churn of staff turnover.
Author’s Offer of An Intriguing Read: I have an article outlining the 115-year-old Scientific Management Methodology that coincidentally walks you through this AI deployment prerequisite…weird but there it is. It’s worth the 7-minute read. Let me know if you want to read about the keys to AI deployment success.
The dealerships that win over the next few years will not be the ones with the mostest, shiniest and newest AI Agents. They will be the ones whose AI makes the humans in the showroom matter more. With the right data prep, business process documentation, and good faith compliance effort as well as staff with the training, deal context, H2H experience, and who are empowered to act on it, will reduce threats to the franchise dealership model from Carvana, Tesla, et al. Clearly, neither vehicle buyers nor salespeople have enjoyed the current, traditional and predominant car sales process. We finally have the tools to build something both parties can trust.

authored by
James Lawrence
Jim Lawrence is an automotive retail systems SME with 25+ years of automotive retail experience and Founder of ImagineMyDreamCar.com (IMDC), DealerEducationFX.com, and DealerEFX.com. IMDC is Jim’s latest award-winning application, an AI-driven lead generator for performance and luxury vehicle retailers. His core competencies include 20+ years working with SaaS-based CRMs, email lead generation, and lead response quality including OEM and regulatory compliance. He holds the Association of Dealership Compliance Officers (ADCO) Dealership Compliance Officer Professional certification and provides Qualified Individual (FTC mandated role) training for ADCO Members.
Get Curated Insights
Content worth the click

