Author: Michael Quigley, Impel Co-Founder & Chief Strategy Officer
Wrapping up the first half of the year is always a moment of reflection for me. Lately I’ve been thinking about where Impel started and where we are now. Today, Impel is likely the largest private verticalized AI company as measured by revenue.
Launching Impel’s AI suite 18 months before ChatGPT captured the public’s zeitgeist, it was clear we had a first mover advantage in our vertical, but I’m still often met with intrigue when I share that statistic.
- How could Impel possibly have over 8,000 customers when automotive dealers, manufacturers, and marketplaces have historically been seen as technology laggards?
- Why has the adoption of Impel’s AI platform in automotive outpaced others in sectors like healthcare, legal, and real estate?
The reality is that 79% of dealers are using AI in their sales process (Source: Car Dealership Guy), and many have advanced far beyond this use case.
The reason why comes down to three vectors with outsized signal in automotive:
1. Query Volume
2. Employee Churn Rate
3. Probabilistic Conversations
These vectors have together led to outsized ROI for our customers and will likely indicate which other verticals are fast followers for AI adoption.
Vector #1: Sky-High Query Volume
Car dealers live and die by leads, and automotive retail is high-touch. The average dealer receives 278 leads per month, of which 166 are qualified sales leads (Foureyes). These leads move through the sales funnel with multiple touchpoints, including everything from financing applications to scheduling test drives. For a rooftop moving hundreds of units, the flood of inbound queries at each of these touchpoints creates an environment where AI adds exponential leverage.
By contrast, most real estate agents on a team will see just 20-30 leads a month and face brutal conversion rates between 0.5-1.2% (National Association of Realtors). The road to a single closed transaction is therefore a slow burn that can take months and hundreds of leads. Fewer queries, longer cycles, and higher emotional stakes mean there’s less room—and less appetite—for automation at scale.
Vector #2: Humans Quit, AI Doesn’t
Employee turnover in auto retail is staggering, with recent data showing annual percentage turnover in the 40s for dealership employees and as high as 67% for sales positions (Cox Automotive). In fact, 73% of US car salespeople leave within two years (Wards Auto).
Now compare that to law, a field where the barriers to entry are high—law school, bar exams, licensing—and the incentive to stick around is built into the profession. The average attorney stays in their role for over 5 years, or about 5x as long as your average car salesperson (BLS).
Dealerships are in a constant state of retraining. Institutional knowledge walks out the door quarterly. Customer follow-ups fall through naturally. Momentum resets every few months. Think 24/7 customer operation staffed by a rotating cast. That’s where AI shines; it doesn’t forget, it doesn’t burn out, and it doesn’t put in two weeks’ notice. AI delivers consistency in an industry where people come and go all too quickly.
Vector #3: Probabilistic Oriented Conversations = LLM Sweet Spot
LLMs are probabilistic. They don’t “know” facts the way a calculator does and instead generate likely responses based on training data. It’s no wonder that LLMs excel at pattern recognition and auto completion but are notoriously bad at math.
The automotive industry plays to LLMs’ strengths since conversations are generally about inventory availability, showroom hours, survey follow-ups, or appointment bookings. Sensitive tasks like payment calculations or final negotiations still rely on the human involvement of dealership personnel. With a focus on response time and less deterministic conversations, automotive is the perfect LLM playground.
Domains like healthcare are a different story. The core use case is highly deterministic by design. Input symptoms, patient history, and lab results—output a diagnosis or prescription in precise quantities. A hallucination here isn’t just a UX issue, but could mean the difference between life and death. When the margin for error is razor thin and the cost of being wrong is so high, probabilistic systems aren’t a viable solution.
In short, automotive lives in the in-between where AI doesn’t have to be ‘Einstein,’ –just ultra responsive, persuasive, and personalized, in order to deliver massive customer value.
Impel was built for this moment, and automotive has proven to be the perfect vertical for AI to flourish; the three vectors above have driven automotive’s outsized acceptance of AI, and its tangible impact on our industry from an ROI perspective. I expect other industries that share these same three characteristics will see vertical AI companies flourish in short order. Many to come.