My first car was a used Pontiac Le Mans. My dad told me not to buy it, and he was right. I bought more car than I could afford, learned the hard way what a few add-ons do to a monthly payment, and eventually traded down to the Volkswagen he’d wanted me to get in the first place.

I was a payment person back then; I could make the number on the sticker. So, I signed! I didn’t think much about what that loan would look like three years in.

A lot of borrowers today are living some version of that same story, and the data tells us lenders need to pay closer attention to it than they have been.

Auto lenders originated a record $211 billion in new loans in the second quarter of this year, according to the Federal Reserve Bank of New York. At the same time, serious auto loan delinquencies climbed to their highest level in roughly 15 years, and that strain isn’t confined to subprime paper. It’s showing up across prime and near-prime tiers, too.

Those two facts sitting next to each other should stop every lender for a minute. Record volume looks like a healthy market. Rising delinquency across credit tiers says something underneath that volume isn’t being priced or structured correctly.

What’s notable about the delinquencies I’m watching is that they’re largely late-stage, not first-payment defaults. These are loans originated in 2023 and 2024 that are now three or four years in and falling 60, 90, 120 days behind (the point in a loan’s life where recovery gets very difficult). A lot of that is economically driven. Vehicle prices climbed hard over the last several years, so when people traded in a car they’d financed long-term, they often rolled negative equity into the next deal.

Add a transmission repair or a job disruption on top of that negative equity, and a borrower with little or nothing to lose by walking away from the loan often does exactly that. Lenders underwriting deals today need to take that lesson from the 2023-2024 vintage and factor it directly into how they’re originating now — not just on FICO, but on debt ratio, on intent to pay, and on what happens to this specific borrower if something goes wrong three years into the loan.

That’s the backdrop against which we ran a survey of 1,500 U.S. consumers this year on how they’re approaching borrowing, and one finding jumped off the page: buying a vehicle is the single leading reason people take out a loan at all, cited by 24.3% of respondents overall, ahead of covering an emergency expense, consolidating debt, or even buying a home. That number climbs to 33.0% among parents with children under 18 and to nearly 30% among borrowers age 35 to 44.

People aren’t borrowing for vehicles because it’s exciting, although I’ll admit it usually is. I’ve bought a lot of cars in my life and I understand the exuberance. They’re borrowing because a car is how most American households get to work, get their kids to school, and keep their lives running. That practical weight is exactly why getting the financing conversation right matters as much as it does.

The survey also surfaced that 47.8% of respondents said the best interest rate matters most when evaluating a loan, while 38.6% pointed to the lowest monthly payment instead. That split is a reminder that “affordability” doesn’t mean the same thing to every borrower. A payment-first mindset usually means someone is stretching to make a vehicle fit inside a tight monthly budget.

A rate-first mindset means someone has more room to negotiate the total cost of the loan. Treating every borrower the same, or worse, treating a whole income bracket as one undifferentiated “bucket,” misses the point. Income is a factor, not a verdict. I’ve seen plenty of prime FICO buyers sitting in lower income tiers. What lenders actually need is the debt ratio, the equity position, and the individual circumstances behind the number, not a category.

That’s also where I’d push back on the instinct to treat a manageable payment and a sustainable loan as the same thing. They aren’t. You can get almost anyone to a payment they can afford this month by stretching the term — 72 months, 84 months — but if the vehicle depreciates faster than the loan amortizes, you’ve just engineered the negative equity that shows up as a late-stage default three years from now.

Lenders have access to the data to avoid that trap like the vehicle’s expected life, its depreciation curve, the borrower’s income and debt ratio, and what happens to all of that if something breaks. The tools to pull that together and analyze it fast now exist. The discipline to use it before approving an 84-month loan on a vehicle that won’t last 84 months is the part that has to improve.

There’s also a fraud and verification piece that’s changed faster than almost anything else in this business over the past year. AI has made it meaningfully easier to produce convincing but fraudulent income documentation, which means lenders can no longer treat a pay stub or bank statement at face value. The same data connectivity that lets us verify a document’s authenticity against actual deposit activity, in near real time, is the connectivity that lets us build a decision around this specific consumer rather than a broad credit category.

Speed matters — nobody wants a borrower sitting at a dealership for hours waiting on an answer — but speed without that underlying analysis is how a lender ends up contributing to the exact delinquency pattern we’re discussing.

Where I think the industry still underserves borrowers is the sequencing of the sale and the financing. A consumer walks into a dealership, picks a vehicle, and only then gets into the real financing conversation in the F&I office, by which point they’ve already made their buying decision and just want the keys. That’s backwards. The alternatives, the different structures, the honest conversation about term, rate, payment and total cost should happen earlier, alongside the vehicle decision, not after it. Dealers are, understandably, focused on selling the car.

Lenders need to bring the financing education into that same moment rather than treating it as a separate, later transaction. Borrowers in our survey told us trust matters enormously in this decision, and trust is built by information offered early, not information delivered after the fact in a room where the car is already sold.

If I had three changes to recommend to auto lenders heading into the rest of this year, they’d be these.

First, underwrite the vehicle and the borrower together. Look at depreciation and equity, not just credit score, before you set the term.

Second, build the capability to tailor deal structures to the individual, not the income bracket or the credit category; the data to do that already sits inside most lending operations.

Third, work with dealer partners to move the financing conversation earlier in the buying journey, so a payment-sensitive borrower understands their real options before they’ve mentally already bought the car.

A record quarter of originations is not, on its own, a sign of a healthy market. A market is healthy when the loans being written today are less likely to be the delinquencies we’re reading about in three years. Consumer data makes it clear that payment-sensitive borrowers aren’t hiding what they need. In fact, they’re telling us plainly.

The lenders who listen, and who build their underwriting and their customer experience around what this data is actually saying, are the ones who will look back on this period as a strong one, not a strained one.

Tom Allanson has more than 25 years of executive leadership experience in fintech and lending. He brings a unique background spanning technology management, entrepreneurship, and enterprise leadership to his role as CEO of defi SOLUTIONS, a pioneer in end-to-end, SaaS loan origination, servicing, and managed servicing solutions.