A store can lose $900 on a used unit before the car ever hits the recon lot. Not because the buyer missed the book. Because the right customer was sitting in the service lounge three weeks earlier, equity-positive, driving a local trade you actually wanted — and nobody connected the dots until that same model showed up at auction with fees, transport, and 14 more unknowns attached.
That is why the recent line from Ken Garff Automotive Group caught my attention. Steve Peterson, the group’s director of data engineering, described the 70-plus-store retailer as “a data company that sells cars” in a Domo announcement covered by Auto Remarketing. I know that phrase can sound like conference-stage perfume. But coming from a family-owned dealer group that has to make payroll, manage grosses, and keep service advisors from revolting, it lands differently.
I think of us as a data company that sells cars.
— Steve Peterson, Ken Garff Automotive Group, via Domo announcement reported by Auto Remarketing
The mistake is thinking this is an IT story
Most dealers do not have a data problem in the literal sense. They have data everywhere. The DMS knows the RO history. The CRM knows the last equity pitch that went nowhere. The desking tool knows the payment. The inventory tool knows aging. The service scheduler knows who is coming in next Thursday. The OEM portal knows lease maturity. The problem is that none of those systems cares whether your used-car manager needs five clean late-model SUVs by Friday.
I have seen this play out at stores from Phoenix to Pittsburgh. The dealer principal pays for more reports, the managers get more dashboards, and the same two things still happen: the used-car department buys too many units from the highest-friction channel, and the service drive keeps handling potential acquisitions like they are just oil changes with a declined cabin filter.
A true data company does not mean everybody stares at charts. It means the store uses information to make a decision sooner than the market does. That is the only advantage that matters. If your data tells you on Tuesday that Mrs. Lopez has a 2021 Grand Cherokee with service history, equity, and a high probability of replacing it, but your process does not trigger a human or automated outreach until next month, you are not data-driven. You are data-adjacent.
The four clocks running against your used-car department
Used-car operators already think in days-to-sale. The sharper stores are starting to think in days-to-acquire. That clock starts before the vehicle is available to the open market. It starts when a service appointment is booked, when a payoff drops below market value, when a customer clicks a trade tool, when a declined repair makes replacement more likely.
Here is the framework I use with stores: every acquisition opportunity has four clocks. If one of them runs too long, the unit usually gets more expensive or disappears.
- Signal clock: How quickly does the store identify that the customer or vehicle is a candidate?
- Contact clock: How quickly does someone make a relevant offer or start a real conversation?
- Appraisal clock: How fast can the store produce a number the customer believes?
- Replacement clock: How cleanly can sales, service, and F&I move the customer into the next vehicle without creating friction?
Most dealerships over-focus on the appraisal clock. They want the perfect number. Fine, but the perfect number delivered four days late loses to a decent number delivered while the customer is still emotionally dealing with a $2,400 service estimate.
Dashboards do not fix broken handoffs
I like good dashboards. I also know a bad handoff can beat a good dashboard every day of the week. If the used-car manager has to beg service for equity customers, the process is already dead. If advisors think sales is going to pounce on their CSI, they will protect their lane. If the BDC is sending generic “we want your car” messages to everyone with a pulse, customers tune it out.
The Ken Garff quote matters because it points to an operating model, not a software stack. A data company that sells cars would not treat service, sales, and inventory as separate kingdoms. It would ask: which customer data should change what we do today?
| Data signal | Typical store behavior | Better operating response |
|---|---|---|
| High-equity customer with upcoming RO | Advisor handles visit normally | Flag before arrival and prepare a trade conversation |
| Customer declined major repair | RO closes and customer leaves | Trigger replacement outreach within hours |
| Local one-owner unit in a hot segment | Wait for customer to raise trade | Proactively give a real buy figure |
| Lease maturity plus service visit | Sales and service both assume the other followed up | Assign one owner and track outcome |
Look, I am not pretending this is easy. Dealer data is messy because dealership life is messy. Customers change phones. Payoffs move. Advisors are busy. Salespeople cherry-pick. Managers override processes when the showroom gets hot. The data does not fully support every shiny promise being made around AI right now, but I would argue the advantage is real when automation is tied to a specific workflow instead of another inbox.
Where AI actually earns its keep
AI in a dealership should not be judged by whether it writes clever emails. It should be judged by whether it compresses the four clocks. Did it identify the right customer sooner? Did it send a timely, relevant message? Did it route the opportunity to the right manager? Did it keep following up without making the store sound desperate?
That is where dealers using tools like AutoRelay have a practical angle, especially in the service lane. The point is not to replace the used-car manager’s eye. The point is to stop depending on memory, sticky notes, and “tell me if you see anything good today” as an acquisition strategy.
Auction buying is still necessary. So are trades from the showroom. But if your cleanest, most knowable inventory source is already visiting your service department and you are not mining it with discipline, you are choosing the more expensive version of the same unit.
Run this acquisition leak test
Pull the last 30 days of customer-pay ROs for vehicles six model years old or newer with fewer than 90,000 miles. Match them against your current used-car stocking needs. Then answer five questions:
- How many of those customers had positive or near-positive equity?
- How many received a specific acquisition or trade message before the RO closed?
- How many got an actual appraisal number?
- How many were assigned to one accountable person?
- How many similar units did you buy from outside channels during the same period?
Now put a dollar figure on the gap. Use this back-of-napkin formula: outside acquisition cost minus internal acquisition cost, plus any gross or recon advantage from owning the history. Even if the answer is only $600 per unit and you missed 10 units, that is real money. More important, it tells you whether you are running a data-driven store or just collecting data while the market beats you to your own customers.
See how AutoRelay helps dealers acquire inventory from their own service drive → getautorelay.com