Dealer risk monitoring that finds a bad dealer in weeks, not after a year of losses
A bad dealer found in weeks, not after a year of losses.
See one case, screen by screen ↓Watching every dealer an indirect auto lender buys contracts from for signs of inflated or fraudulent deals and early defaults, so a problem dealer is found and acted on early. Dealer Risk Monitoring scores every active dealer each night on six signals, against its own history and dealers like it, and drills from the score to the contracts behind it.
How to choose AI solutions for lending →
Updated
In indirect lending you never meet the borrower — every contract is written by a dealer, and a small number of dealers write most of the trouble: vehicles booked with options they do not have, service contracts and GAP packed into the payment, paystubs from employers that do not exist, and the first payment that never comes.
The signals are all there, but on different desks. Funding sees the stipulation pends, the fraud team sees the failed income checks, collections sees the early defaults, compliance sees the complaints. Credit risk puts them together by dealer in the quarterly review — and early payment defaults only surface 60 to 180 days after funding. By then the dealer has written hundreds more deals, and without the sample deals that prove it, the relationship manager pushes back and the repurchase clause goes unused.
Estimated split for one problem dealer, in days — by quarterly review, and with the solution. First payments still have to come due either way.
Seven specialist agents score every dealer nightly and find patterns across its deals; the analyst decides.
Coppervale Bank buys contracts from 412 dealers in Texas, Oklahoma and Louisiana. On Monday 12 October one of them, Brazos Bend Auto Plaza in Houston, scores 91. Here is what the dealer risk analyst sees, screen by screen, in the working solution.
Dana opens a map, not a report: 412 dealers as dots, each sized by contracts funded in 90 days and coloured by risk. One red dot in Houston stands out — Brazos Bend Auto Plaza, 91, with an alert. The side rail shows its score crossing the 80 line, early defaults at 6.1% against 1.2% for peers, add-ons on 87% of deals against 38%.
“Dealer risk scorer scored 412 dealers at 05:12 · 5 alerts open.”
She zooms to Houston and replays the network week by week. In the week of 23 August Brazos Bend is an amber 55 on 101 contracts; by October it is red. The timeline marks the pattern change on 24 August and the alert on 2 October, when the score crossed 80.
A single deal review sees one $2,000 down payment. Across the dealer’s 148 contracts since 14 July, the same figure repeats seventeen times — with the same two employers, one employer phone, (281) 555-0164, the position “route driver”, and stated incomes within $525 of each other. Four have already missed the first payment, $439,720 is financed, and two more are waiting to fund.
The pattern finder never calls a pattern fraud — it shows the deals for a person to verify.
The seventeen open in the deals table under a “Same $2,000 down” filter: vehicle, amount financed, cash down, book, advance, add-ons, the employer check and the first payment for each. Ridgeway Logistics and Lone Star Fleet Services, again and again — and “Could not verify” under every one.
Four agents re-read the 148 contracts, follow every first payment and join complaints and fraud referrals. The score breaks down as employment 24 of 25, early default 23, value 17, add-ons 14, complaints 8 and pends 5, explained against 214 independent dealers — while monthly volume ran about 24 a month to June, then 31 in July, 46 in August and 61 in September.
The phone line is not only at Brazos Bend. Westpark Auto Sales, also in Houston, sent five applications naming the same two employers since 14 July. Ridgeway Logistics has no business at 4400 Almeda-Genoa Rd and is not registered with the state; Lone Star Fleet Services was registered on 3 August with the same registered agent. The number answers as a voicemail for either company.
The review pack samples twelve deals across the signals and checks each one where it can be proven: the employer called and the state register searched, the VIN decoded against the build sheet, the product menu read. Four paystubs are falsified, five deals carry options not on the car, three menus are unsigned. Contracts BB-0918, BB-0931, BB-0947 and BB-0962 — $97,640 — are eligible for repurchase under §7.2 of the dealer agreement.
“About $812,000 net loss on the 263 open contracts over 12 months · $214,000 for a dealer its size.”
The recommendation is the least action that stops the pattern: employment and income verified by the bank on every deal, advance capped at 115% of book on VIN-decoded options, service contract and GAP each on a signed product menu, the repurchase demand, and a review in 30 days. The notice to the dealer is drafted and sent only when signed.
Termination would strand Curtis Bell’s relationship before the 30-day test.
Dana signs the restriction with her password and the meaning of her signature: “I reviewed the pack and the deals behind it”. A restriction needs a second signature, so it goes to Russell Teague, Head of Auto Credit Risk, before the notice is sent.
Russell co-signs from his phone; the notice goes to Brazos Bend and Curtis Bell is briefed. The action log shows it at the top: Restrict, today, a review on Thu 12 Nov and $97,640 repurchase demanded — ten days from alert to signed action, five weeks after the pattern changed.
“Found in 5 weeks, not at the loss review.”
The same dealer, seen by the five people who carry it — the analyst, the credit risk head, the fraud manager, the relationship manager and the head of auto finance.
Seven specialist agents score every dealer nightly and find patterns across its deals; the analyst decides.
Reads every funded contract each night: advance to book against VIN-decoded options, add-on products, stipulation pends and employment checks — and attributes each to its dealer.
Follows each contract’s first six payments and computes first-payment and early default by dealer and by month funded, against the dealer’s peers.
Ties borrower complaints, fraud referrals and shared employers, phone numbers and addresses on applications to dealers, and summarises complaints by theme.
Looks across each dealer’s deals from the last 90 days for details that repeat far more often than at dealers like it — the same cash down, stated income band, employer, employer phone, references or position.
Scores every dealer against its own history and dealers like it, explains the score as points by signal with the contracts behind each, and raises an alert when the pattern changes.
Samples deals behind each signal, verifies them at the source — the employer at an independently found number, the state register, the build sheet, the signed product menu — and estimates the exposure on the open contracts.
Drafts a watch, restrict or terminate recommendation with conditions on new deals, repurchase demands under the dealer agreement, the dealer notice and the relationship manager’s talking points.
Decides; Head of Auto Credit Risk co-signs a restriction. The agents propose; a named person decides.
Ask in plain words — or tell it what to pend or hold. Every answer points to the contracts and documents behind it.
Why did Brazos Bend’s score jump to 91?
Brazos Bend Auto Plaza went from 63 to 91 in 30 days. Four signals moved together after 24 August, while volume went from 24 contracts a month to 61: employment not verified on 22 of 71 (31%, peers 6%), 9 first payments missed (6.1%, peers 1.2%), advance at 131% of book with 34 deals with options not on the VIN (peers 112%), and service contract plus GAP on 87% (peers 38%).
Which deals show falsified income?
Four, confirmed in the review pack — each employer called and searched on the state register: BB-0918, BB-0931, BB-0947 and BB-0962. Together $97,640, eligible for repurchase under §7.2. Another 18 could not be verified.
Which other dealers use the same employers?
Westpark Auto Sales in Houston: 5 applications since 14 July name Ridgeway Logistics or Lone Star Fleet Services, and both employers list the same phone line, (281) 555-0164. Westpark’s score rose from 41 to 64 in 30 days; 4 of the 5 are funded. No other dealer names either employer.
Pend the 3 Brazos Bend contracts waiting to fund until income is verified
Done. Three Brazos Bend contracts are waiting to fund — $71,860 in all. Two name Ridgeway Logistics or Lone Star Fleet Services and put down exactly $2,000. “Pend 3 contracts until employment is verified” is added to the restriction; they fund once the funding desk reaches the employer.
13 screens from the working solution, on its sample data. Pick one to see it large.
How dealers are scored, when an alert is raised and who may act — changed in settings, versioned and on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| Signal weights | Income 25 · early default 25 · value 20 · add-ons 15 · complaints 10 · pends 5 | Adjust each, total 100 |
| Raise an alert when | 15 points in 30 days | 10 · 15 · 20 points in 30 days |
| High-risk line (reviewed within 10 business days) | 80 | 75 · 80 · 85 |
| Dealers compared with | Franchise, independent, and by 90-day volume | Franchise and independent · All dealers together |
| Minimum volume to score | 10 contracts in 90 days | 5 · 10 · 20 contracts in 90 days |
| Advance-to-book tolerance | 115% of book | 110% · 115% · 120% of book |
| Who may restrict | Analyst + Head of Auto Credit Risk | Analyst + Fraud Manager · Analyst alone |
| Review after a restriction | 30 days · limit 3% | 60 days · limit 3% · 30 days · limit 2% |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample dealer data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (Point Predictive, 2025 Auto Lending Fraud Trends Report — dealer risk up to 500%, up to 70% of early defaults show application fraud; California Attorney General, $694M multistate settlement, 17 Sep 2026). People, dealers, employers and the bank named on this page are fictional — characters and sample data in the working solution.
Watching every dealer an indirect auto lender buys contracts from for signs of inflated or fraudulent deals and early defaults, so a problem dealer is found and acted on early. Dealer Risk Monitoring scores every active dealer each night on six signals, against its own history and dealers like it, and drills from the score to the contracts behind it.
Employment and income, early payment default, vehicle value against book, add-on products, borrower complaints and funding pends. The weights are settings — 25, 25, 20, 15, 10 and 5 out of 100 in the working solution.
The deal signal reader compares the options booked on each contract with the VIN decode and records the advance to book, and reads the service contract and GAP on every deal. A dealer whose deals carry options the build sheet does not show, or both products far more often than its peers, moves up the ranking — with the deals listed.
No. The pattern finder shows repeated details and the deals that carry them; it never calls a pattern fraud. The review pack verifies sampled deals at the source, and only a deal with a document on file counts as proven. The agents never restrict, terminate, contact a dealer or a borrower, or change a funding decision.
The dealer risk analyst signs a watch. A restriction needs the Head of Auto Credit Risk’s co-signature, and a termination also needs the Head of Auto Finance. Each signature records the meaning, and the notice to the dealer is sent only when every signature is in.
Yes. When the review pack proves false information on a contract, the contract is listed as eligible for repurchase under the dealer agreement, and the demand goes into the recommended action and the dealer notice for a person to sign.
Your loan system, the funding desk’s pends and stipulations, employment and income checks, servicing payments, the complaints system, VIN decode and book values, and your dealer agreements.
The Agentic Solution Engine builds and deploys it from your requirements — your credit policy, dealer agreement and a sample of funded contracts — and it goes live once every quality gate has passed. We will walk you through it on your own dealers first.
We’ll run Dealer Risk Monitoring on a sample of your own funded contracts.