Cox Automotive has the US market closing 2026 at about 15.8 million units, down 2.9% from 2025, with the first half already running 3.6% behind last year. The monthly pace has held up better than the headline suggests — June came in at a 16.5 million SAAR, up 7.2% year over year — but the shape of the year is clear enough. Volume is not going to rescue anybody.
I want to be careful about how that gets read, because "down 2.9%" gets reported as a bad year and it isn't one. Sixteen million cars is a lot of cars. What's actually happening is narrower and more interesting: the market has stopped growing on its own, and affordability is doing the sorting.
Flat markets don't reduce demand. They redistribute it.
Cox's own framing is that affordability remains the central constraint — elevated rates, higher costs for essentials, tighter household budgets — while strong equity markets and accumulated wealth prop up the top of the demand curve. That is not a market with less demand in it. That is a market where demand has become selective, and where the customer who would have replaced a vehicle at 48 months now replaces it at 60, or 72, or keeps it and services it.
The first-half scoreboard shows how unevenly that lands. GM was down 7.2%, with Buick and Cadillac each off more than 20%. Ford was down 10.3%. Tesla down 14.6%. Meanwhile Hyundai Motor Group picked up seven-tenths of a share point, Stellantis put volume up 4.8%, and Toyota's Q2 ran 18.8% ahead of its Q1. Same market, wildly different outcomes.
When a market grows, weak retention is invisible — new traffic covers it. When a market flattens, retention is the whole game, because the only customers that are definitely still there are the ones you already sold.
The job changes, not the goal
In a growth year, a dealership's marketing job is roughly "capture your share of the people walking in." In a year like this one, it's three different jobs:
- Know who in your database is actually in-market. Not everyone. The 4% who are moving. Equity position, service history, mileage against loan term, and behavioural signals will tell you, if anyone is reading them.
- Get to them before a third-party lead provider does. You already paid to acquire these customers once. Paying a portal to rent them back is the most expensive way to lose an argument.
- Defend the service lane like it's the sales floor. Because it is. Cox's 2026 ownership study puts the dealership share of all service visits at 29%, down from 33% in 2018, and buyers who service where they bought are 74% likely to repurchase versus 44% who don't.
That third point is the one I'd put on the wall. A customer you keep in the service drive is not a fixed-ops win that happens to be nice. It is your next new-vehicle sale, statistically, at nearly double the odds.
What we're telling our own dealers
Nothing clever. Three things, in order:
- Stop buying the same customer twice. Before you increase conquest spend, find out what percentage of your service-active customers are getting any structured communication at all. In most stores the honest answer is "the ones the OEM emails."
- Fix the measurement before the campaign. If sales, service, parts and inventory are still four separate reports that disagree with each other, you cannot tell a good month from a lucky one. That is the entire reason we built Pie.
- Treat a flat market as a planning assumption, not a surprise. Budget for 15.8. If it comes in at 16.5, you have a good year instead of an emergency.
Sixteen years in this business has taught me that dealers don't lose in soft markets because the market is soft. They lose because the habits that worked in a growth year — spend more, chase more, worry about the database later — stop working, and it takes most of a year to notice.
This is a good year to notice early.
Figures: Cox Automotive June 2026 US auto sales forecast and the Cox Automotive dealership ownership study, fielded September–October 2025 and published April 2026.
