Stop Optimizing the Car

In our May issue, Two Turns and Out, we laid out the problem: the car now breaks down faster than the loan, and the old buy-here-pay-here wisdom — turn a unit until the wheels fall off — is finished. This issue is about what the best operators do once they accept that.

If the asset is the thing that keeps failing you, you can’t underwrite it the way you used to — and you can’t underwrite the borrower the way you used to either. The operators whose books are still performing have stopped treating underwriting as a single credit decision made once, at the sale. They underwrite two things at the same time, and they keep underwriting both for the life of the loan: the durability of the car, and the durability of the relationship with the person driving it. Call it multi-factor underwriting. The credit history, the down payment, the job history — that’s table stakes. The questions that actually decide whether this loan performs are narrower and harder: will this specific car survive the term, can this borrower afford to keep it on the road, and have you given them a reason to keep paying the note instead of walking away.

That’s a different business than the one most of the market still runs. It’s colder, not softer — a clearer read of where the money actually is. The performing, returning borrower is the only appreciating thing on the lot; everything else depreciates on a schedule. The rest of this issue is how the best operators underwrite for that, what it takes to reward the customer without kidding yourself, and the squeeze that’s forcing everyone else to catch up.

The Squeeze That Forces It

Start with why this is happening now, because the strategy only makes sense against the backdrop. The demand isn’t going anywhere — but the affordable car is. Sub-$15,000 used inventory was down about 26% year over year this August, per Cox Automotive — just 15.1% of the market, under a month’s supply. The subprime borrower is still there — subprime hit its highest fourth-quarter financing share since 2021, per Experian — and straining, with sixty-plus-day subprime delinquency on auto ABS printing a record 6.9% in January, per Fitch. But keep those figures in proportion: they track the securitized, Carvana-and-captive end of subprime, not the lot down the road. At the buy-here-pay-here level the numbers are far harsher — in our experience, delinquency and default run 30% and north.

The credit box is compounding it. Over the past year the big subprime and nonprime lenders have raised their income-to-payment bars — Ally, for one, put 43% of its auto originations in its highest credit tier in the fourth quarter. Even buy-here-pay-here is drifting up-market: the Fed found dealers cut their deep-subprime share from around 70% of loans in 2018 to just above half by 2025. The result cascades down the price ladder. The borrower who’d have financed a 2021 model can’t, so they drop to a 2017; the one who’d have taken the 2017 drops again — the whole subprime market gets pushed down the vehicle-age ladder at once, into exactly the older, breakdown-prone iron this model has to keep running.

And the exit a lot of operators counted on is narrowing. The older idea was that buy-here-pay-here could be a bridge — get the borrower into a car, let them build a record, hand them off to a bank at a lower rate. That path is mostly closed now. Banks have tightened, the warehouse credit facilities that fund this paper are pulling back, and a rate cut or two from the Fed does little for the cost of funds on a deep-subprime book. For most operators the borrower they sign today isn’t passing through to someone else’s balance sheet — they’re staying on yours, at a high cost of money, for the life of the loan.

You can’t price your way out, you can’t refinance your borrower out, and you can’t count on the next cheap car showing up.

Put it together and the message to the operator is simple. What you can control is two things: make the car you sold last the loan, and make the customer you signed worth keeping. That’s the whole game now — and it’s two separate underwriting problems, not one.

Underwrite the Car, Not Just the Credit

Here’s the trap I watch dealers fall into. Volume’s slow, the cheap cars are gone, so they reach up-market — nicer, newer, flashier iron — and tell themselves they’re buying quality. I had this exact call with a dealer not long ago. Same town, same customer base, but he’d swapped his lot over to higher-end units — a Nissan Titan with the big V8, softly-used Lexus, that kind of thing — and shaved his rate to make the payment pencil. I told him what I’ll tell you: that’s lipstick on a pig. You multiplied your cost basis in the asset and cut your rate to cover it, on the same borrower who was already struggling. If he defaults, your loss on a $15,000 ACV is a different animal than your loss on a $6,000 one. You didn’t reduce risk. You dressed it up and called it a strategy.

Underwriting the car means asking whether this specific vehicle survives the term — and what it costs to keep it alive if it limps. A boring lot of reliable, common sedans and small SUVs — the kind of car anybody can work on, where parts are cheap and everywhere — will out-earn a row of lifted trucks and European badges every time. And there’s a factor almost nobody prices in: insurance. Two cars at the same monthly payment are not the same deal if one costs the borrower $300 a month to insure and the other $150. The expensive one quietly eats the budget you were counting on to get paid. If you walk a borrower into a car they can afford to buy but can’t afford to insure or repair, you didn’t do them — or yourself — any favor. That’s what multi-factor underwriting means on the acquisition side: the repair cost, the parts pipeline, and the insurance bill are underwriting inputs, not afterthoughts.

I’d rather have a lot full of Subaru Foresters I know will survive the loan than a row of lifted trucks that look good at auction.

The sharper operators are creating affordability without stretching the loan to get there. Instead of tacking on term — the 72- and 84-month paper blowing up across the broader subprime market right now, full of four- and five-year-old cars that can’t outlast the note — they’re taking it out of the markup. One book I watched had its average balance at origination come down about 15% that way. You won’t see 72- or 84-month terms at the buy-here-pay-here level anyway; the rule of thumb I give people is that forty months is the sweet spot, maybe thirty-nine for the optics of a three-handle. Anything north of fifty is trouble, and anything north of sixty is a dud. Past that, you’re betting the car outlives the loan, and in this vintage it won’t.

This is also where you need the discipline to let an asset go. A squealing brake job on a performing borrower? Finance it, keep them running. A blown motor? That’s the moment you stop. You don’t drop ten grand on a crate engine to save one deal. You take those dollars, treat them as the borrower’s next down payment, and solve for zero on that first turn — break even on the deal instead of chasing a profit: sell the car, collect the down, run the note, send the tired unit to auction, take your lumps, and come out roughly whole. You look for your profit on the second and third turn, with a customer who now trusts you. Underwrite the car honestly enough to know when it’s earned its exit.

Turn the Customer, Not the Car

The other half of the model is the borrower — not their score at the sale, but their behavior over the life of the loan. The whole point of keeping the car running is keeping the customer paying, because a borrower who believes the car’s about to die has already half-decided to stop paying for it. So when a payer calls in with a problem they can’t cover, the move isn’t to prep the repo. It’s to do the small repair and tack it on as a side note — a separate few hundred dollars, often stretched over the remaining term at no added rate. The goodwill is worth more than the margin you gave up on the brakes.

It starts before the first payment is ever due, at the credit desk, and it starts by moving past the score. The score that once drove the decision no longer does; what carries weight now is a deeper read of the borrower’s credit history and how cash actually moves through the household. The sharp operators don’t stop at rent’s due on the first — everyone knows that. They map the household’s real cash flow: not just income, but the last few months of unplanned hits — the dental bill, the blown tire — and how the borrower actually covered them, because that’s what tells you whether a car payment survives the next surprise. Then they set the payment schedule to how the borrower truly gets paid, not to the lot’s calendar. Just as important is setting the terms of the relationship out loud at signing: we’ll work with you, but you have to call us, and it may not be news we want to hear. The minute the phone goes quiet is the minute a workout becomes a repo. The operators who learn early who tends to go silent can get ahead of trouble before it becomes a loss — the answer you never want is I can’t pay, come get it, because a borrower who’s reached that point has already quit on the deal, and the car he’s handing back is usually the worse for it.

The same instinct shows up in how the best operators collect. Instead of one monthly payment a deep-subprime borrower can’t protect, they take small, frequent bites — weekly, biweekly, in a few cases effectively by the day. The Fed put numbers on how different this already is: in its May 2026 study, roughly 14% of subprime buy-here-pay-here balances sat on weekly or biweekly schedules, against almost none for traditional lenders, where nearly every loan is monthly. Be precise about what that buys, though. More due dates mechanically means more chances to miss one; the Fed’s own read is that the real edge is on the lender’s side — steadier cash flow, and an early-warning system that surfaces a borrower in trouble in days instead of at month-end.

More due dates mean you hear about trouble in days — not at month-end, when the car’s already gone.

Service Is the Retention Engine

Keeping the car running is the other half of the model, and the best operators have turned it into a system instead of a favor. There’s a buy-here-pay-here group out in the rural South built entirely around it — small-town lots feeding a big in-house reconditioning and parts operation, with their own salvage yard so parts are cheap and the car’s back on the road in days, not weeks. The service bay isn’t a cost center; it’s the retention engine. When you own the paper, a breakdown isn’t a warranty headache — it’s a direct hit to a live receivable. A pure lender can’t keep those cars alive. An operator with the wrenches can.

But the deeper idea isn’t giving away oil changes. It’s turning maintenance from an unpredictable expense the customer dreads into a scheduled reason to come back — and operators do it three ways, each smarter than the last. The weakest is the calendar reminder: it’s been six months, come get your oil changed. Nobody listens to their dealer’s email.

The stronger version is contractual — sell the service up front so the customer is already committed to coming back. A tiered plan that bakes a set number of oil changes, roadside assistance, and mechanical coverage into the deal writes several future visits in before the customer ever drives off. Every visit is a touchpoint — a chance to put eyes on the car and catch the small problem before it becomes a repossession.

The cheap oil change isn’t the product. The visit is.

The frontier lets the car itself call the customer in. Instead of guessing from the calendar, connected-car telematics platforms read the vehicle’s actual mileage, its past repair orders, and the factory service schedule, then ping the owner to book the work exactly when it’s due — not a month early, not after the breakdown. Newer connected-car systems go further, reading fault codes and condition data to catch a problem before the engine ever complains. A handful of buy-here-pay-here operators are wiring the same GPS units they once used only to find and disable a car into their payment apps, so a borrower logging in to pay gets a service reminder and a coupon for the shop. It’s early, and worth being honest that it rides on the same hardware built to repossess — but the signal has flipped from it’s been six months since Dave bought a car to this car is actually due, and that’s a far better conversation.

Here’s why this lands harder in buy-here-pay-here than anywhere else. For a franchise store, service retention is a fixed-operations revenue play — nice to have. For the operator who holds the loan, it’s the whole ballgame: a car that stays running is a borrower who stays current. Service retention feeds vehicle uptime, uptime feeds payment continuity, and payment continuity feeds the repeat sale. The service relationship doesn’t sit beside the receivable. It protects it.

Reward the Relationship — Carefully

If the performing customer is the asset, the obvious question is how you reward them for staying.

This is harder than it sounds. The obvious reward — a lower rate for a borrower who’s proven he pays — does exist, but only at the margins. A handful of buy-here-pay-here operators have built real performance-based step-downs, cutting the rate after stretches of on-time payments and letting those cuts stack up over time.

It stays rare, though. There’s no real industry playbook for rewarding good payers with cheaper credit — and the economics tell you why. Deep-subprime runs on a high rate. By the time a borrower’s earned his way to materially better terms, he can usually get cheaper money elsewhere.

So the problem isn’t just how to reward a good customer. It’s how to reward him without blowing up the economics that made the account work in the first place.

That’s the paradox at the center of this. The reward operators used to lean on was credit-building — reporting on-time payments to the bureaus so the borrower’s score recovers. It had its moment: buy-here-pay-here lots historically held their own paper and often didn’t bother, so Equifax dropped its account minimum for small dealers back in 2015 specifically to pull more of them in, and the COVID-era wave of cheap originations briefly made credit-building an easier sell. But the ground has shifted. Reporting cuts both ways, and the tension is built in: successful credit-building makes your best customer more financeable somewhere else — rebuild their score far enough and the credit union that couldn’t touch them two years ago can suddenly refinance them away. So the sharper lots have quietly stopped leaning on it — reporting earns the operator little now, and many no longer report the good, while some have stopped bothering with the bad either. The customer stays put regardless: with credit tightened above them, that borrower is a lifer again — call it a two-year runway before Fed policy changes the math. The real retention comes from the relationship and the next car, not the score.

You can’t sell the same customer seven cars at 22% every time. If he’s proven he pays, he has to be rewarded for it.

A few operators are getting more creative, and it’s worth knowing the ceiling before you copy them. One large subprime lender has taken a more bank-like approach, offering its auto-finance customers a Visa card that earns enhanced cash back on gas, repairs, and other automotive spending and can even be used toward a vehicle down payment — a loyalty layer wrapped around the auto-finance relationship. Which is the point: you can act a little like a bank, but you can’t be one, and you can’t pretend to be. One major subprime auto lender just reached a roughly $700 million multistate settlement after regulators alleged it put consumers into loans they couldn’t afford — loans some states described as designed to fail; the lender denied wrongdoing. Selling ‘we’ll help your credit’ while writing paper the borrower can’t survive is what draws that kind of enforcement. A reward program is not a shield. And the softer perks have limits too — the free-oil-change loyalty play only reaches the customer who still shows up, and the borrower who’s behind is the last person walking into the lot that wants to repo him.

So what actually works? The reward has to be structured, modest, and tied to behavior you can see. The operators already testing it reduce the rate in stages as a customer stacks consecutive on-time payments. The broader principle matters more than the exact formula: give the proven payer something tangible — a better rate on the next deal, a modest price concession, discounted service, a path into better paper — rather than treating the seven-time customer exactly like the stranger who walked in yesterday. None of it is sexy, and that’s the point. The operators who survive this cycle won’t be the ones who found a cleverer way to squeeze the asset — it’s a depreciating liability that breaks on schedule. They’ll be the ones who underwrote the car and the customer together, kept both alive, and gave the good borrower a reason to come back. Turn the car when you have to. But build the business around turning the customer.