Gross profit per unit at used-car dealer groups is sliding back down toward pre-pandemic levels. Overhead, labor, utilities, software, is not sliding anywhere; it’s sitting at record highs. That gap, not any single line item you can point to in a board deck, is the actual 2026 margin story, and it’s the one a COO has to defend when someone with an EBITDA multiple in their head asks why the numbers look worse than last year.
What’s actually driving margin compression in 2026
For three years, dealer groups got a strange kind of cover. Constrained new-vehicle supply pushed transaction prices up, gross profit per unit ballooned well past historical norms, and a lot of operational inefficiency got quietly absorbed by that fat margin. Nobody was auditing cost per transaction too closely when every unit was throwing off outsized gross.
That cover is gone. Bradyware’s 2026 dealership trends brief describes it plainly: gross profit per unit is normalizing back down to pre-pandemic levels, while overhead costs, labor, utilities, software, have stayed at record highs. The two lines used to run in the same direction. Now they’ve decoupled, and the space between them is where margin used to live.
This is why “just sell more units” doesn’t fix the problem the way it used to. Volume without a lower cost base just multiplies a thinner margin. The fix has to come from the overhead side, and specifically from the parts of overhead that scale with headcount and manual work rather than with unit count.
How much did net floorplan expense per vehicle actually rise
One line item inside that squeeze has a hard number behind it. Per Harney Partners’ floor-plan financing analysis, net floorplan expense per vehicle rose by roughly 39 percent, about $139 per unit, in Q2 2025. Over the same year, new-vehicle days supply climbed from about 71 days to 88 days (November 2024 to November 2025), as inventory built up faster than it turned.
Those two numbers are connected. A car that sits longer accrues more floorplan interest before it sells, and more of the fleet was sitting longer. That’s a direct, defensible line from days supply to floorplan cost, and it’s the kind of number a board will accept without asking you to prove causation, because the mechanism is obvious to anyone who has personally guaranteed a floorplan line.
Floorplan expense is a real and growing piece of the squeeze, but it’s not the whole story, and treating it as the whole story is a mistake we see COOs make. It’s one visible symptom of a broader pattern: costs tied to how long a vehicle or a deal sits unresolved are rising across the board, not just at the lender relationship. For a deeper breakdown of the floorplan piece specifically, see Floorplan Interest Per Unit: the Line Item Quietly Eating 2026 Margins.
Why overhead won’t come back down on its own
Labor, utilities, and software costs don’t normalize the way gross profit per unit does, because none of them are priced off vehicle transaction volume. Utility rates and software subscriptions move independently of how many units you sell this month. Labor is the sticky one: title clerks, F&I back-office staff, and CIT (contracts in transit) processors are running the same manual, document-heavy workflows they ran in 2021, except now every hour of that work is a larger share of a shrinking gross.
That’s the actual mechanism behind margin compression, not a single bad line item, but a structural mismatch: revenue-side costs (floorplan, wholesale, acquisition) that move with market conditions, sitting next to headcount-side costs that move with document volume and don’t shrink just because gross profit per unit did. For the fuller picture of why gross profit per unit specifically won’t bounce back to 2021-2022 levels, see Gross Profit Per Unit Normalization: Why 2026 Feels Different.
Here’s the shape of it as two converging lines, which is the version that tends to land in a board meeting better than a paragraph of explanation:
The first line is gross profit per unit: the pandemic spike, then the slide back toward pre-pandemic territory. The second is overhead cost per unit: flat, then climbing, never giving back the ground it gained. The shapes in this chart are illustrative, drawn to match the direction both sources describe, not exact reported figures for any one dealer group. The point isn’t the precise curve; it’s that the lines cross, and once they cross, cost per transaction stops being a background metric and becomes the number the whole P&L turns on. If you want the transaction-level version of that number, Cost Per Transaction: a Realistic Benchmark for Used-Car Operations walks through what a realistic per-deal cost looks like once you account for labor, not just software.
Why this matters more if you’re PE-backed
If your dealer group is owned or backed by private equity, this squeeze lands differently than it would at a family-owned rooftop. A board evaluating an EBITDA multiple doesn’t respond the same way to every kind of pitch. A defensible cost-avoidance story, this specific line item shrinks by this specific amount, clears budget review faster than an unproven revenue-growth pitch, because the board can underwrite a cost reduction with much more confidence than it can underwrite a sales projection.
That shapes how an operations fix needs to be framed internally. “We’re going to grow throughput” is a promise. “We’re going to cut cost per transaction from X to Y because the review queue is the bottleneck, not demand” is a claim you can back with a before/after number. The second framing is the one that survives a board meeting.
The pressure is not abstract right now, either. Broader labor-market anxiety is already reaching dealer groups: PE-backed portfolio companies across sectors filed WARN notices affecting close to 13,000 workers between January and May 2026, and many used-car dealer groups are now PE-owned or PE-adjacent, so that drumbeat is not background noise to a title clerk or F&I processor watching the news. It’s a reason ops leaders need a real plan for absorbing volume growth without headcount growth, not just a reason to freeze hiring and hope. For how that pressure plays out specifically in staffing decisions, see PE-Owned Dealer Groups and the 2026 Headcount Pressure.
What actually moves the number
If overhead is the fixed side of the equation and it’s mostly labor tied to manual document work, the honest lever is reducing how much labor a transaction requires without reducing accuracy. That’s a narrower claim than “adopt AI,” and it’s the one that survives scrutiny.
We think about it in three buckets: direct labor, unlocked capacity, and error reduction. In one document-heavy transaction-processing operation we’ve studied closely, a review team went from 12 people to 6 (the other half redeployed, not laid off) after review time dropped from around 20 minutes to 1-2 minutes per case and the error rate fell from about 7% to about 1%. About half of deals became auto-approved outright, and roughly 70% of total volume ended up AI-managed end to end, including overnight hours when no one was on shift.
Our rule of thumb: it’s worth building a process like that when labor cost, unlocked capacity, and error/leakage together add up to $1.2 million or more a year, and the automation itself should cost no more than about 20% of the value it captures. That threshold matters because it’s the difference between a defensible cost-avoidance story a board will fund and a pilot that stalls because nobody can prove it paid for itself. It’s also worth stress-testing against what due diligence teams actually flag when a dealer group changes hands; see Dealer Group M&A Due Diligence: the Operational Red Flags Buyers Actually Find for what an acquirer’s ops team looks for before they’ll sign off on a valuation.
None of this makes floorplan interest, days supply, or acquisition cost go away. Those are market-driven and mostly outside an ops leader’s control quarter to quarter.
Key insight
What's inside your control is cost per transaction on the document-heavy side of the business: title work, deal jacket review, CIT processing, F&I back-office. That's the lever that doesn't depend on used-vehicle pricing coming back.
FAQ
What’s driving margin compression at dealer groups in 2026? Gross profit per unit is normalizing back down to pre-pandemic levels while overhead costs, labor, utilities, software, remain at record highs. The two used to move together during the 2021-2022 supply crunch; now they’ve decoupled, and the gap between them is what’s eating margin.
How much did net floorplan expense per vehicle rise? Roughly 39 percent, about $139 per unit, in Q2 2025, according to Harney Partners’ floor-plan financing analysis. Over the same year, new-vehicle days supply rose from about 71 days to 88 days, as more inventory sat longer before it sold.
Why does this matter more for PE-backed dealer groups specifically? Boards evaluating EBITDA multiples respond more strongly to a defensible cost-avoidance story, this specific cost shrinks by this specific amount, than to an unproven revenue-growth pitch. That shapes how an ops leader needs to frame a fix to actually get it funded, and it matters more now given the broader wave of PE-driven headcount pressure moving through the industry in 2026.
If your document-heavy back office (title, deal jacket, CIT) is the part of overhead that isn’t shrinking, Deskflow is built around exactly that three-bucket math: labor, capacity, and error reduction, applied to the transactions your team is already processing.