Warranty Claims Are Your Most Rules-Driven Process. Here Is Why They Are Still Manual

The process that should have been automated a decade ago

Walk a dealership group's fixed operations and ask which process is most rules-driven. Everyone says warranty. The labor operation codes are published. The time allowances are published. The documentation requirements are published. The claim goes to one payer on a known schedule, and the payer tells you in writing when it disagrees. On paper this is the cleanest automation candidate in the building.

Then look at who does it. A person. Often one person per franchise, sometimes one person for four rooftops, working a queue in the DMS and a manufacturer portal side by side, with a binder or a shared drive of policy bulletins that she maintains herself. She has been there eleven years. When she takes two weeks off, the department feels it.

Every vendor that has pitched your group reads that gap as a software problem. It is not. The reason warranty stays manual is structural, and until you can name it, you will keep buying tools that automate the part that was never the constraint.

The asymmetry that keeps a human in the seat

Warranty has two failure modes and they are not priced the same.

If a claim is submitted wrong, the manufacturer pays it, then audits it later and takes the money back. That is a chargeback. It arrives after the money has been recognized, after the technician has been paid, sometimes after the fiscal year has closed, and it arrives in a batch. If a claim is submitted slowly, the manufacturer pays it late. That is a receivable. It sits on a schedule, it annoys the controller, and eventually it converts to cash.

One of those failure modes threatens the P&L. The other threatens a line on a schedule. Given that spread, every rational incentive in the store points at careful humans reviewing every claim, and no incentive at all points at speed. That is the answer to "why is this still manual," and it is a correct answer, not a legacy one. The group is not behind. It is responding accurately to the risk it faces.

What makes the asymmetry sharp is that the rules change without notice and they change per manufacturer. Hyundai, Ford and Stellantis brands do not share documentation standards, photo requirements, prior-authorization thresholds, or what counts as an acceptable cause-correction-condition narrative. A group with six franchises is not running one warranty process. It is running six, and the only thing holding them together is that each one lives in a specific person's head. On the Fixed Ops Round Table podcast, David Pickett of MileOne described consolidating twenty-one warranty administrators down to five in a corporate office, and named the same brands as the hard ones. His warning about the staffing risk was blunt: "two months of warranty claims backing up can turn into a $100,000 audit in no time."

That is the shape of the problem. Not a bad process. A concentrated one.

Where the money is actually leaking

Here is what most groups have never measured, because the person who could measure it is the same person doing the work.

State law sets the real clock on warranty, not any internal service department target. Every state now has a retail warranty reimbursement statute, and the provisions that matter operationally are remarkably consistent. In Minnesota, claims must be approved or disapproved within thirty days, anything not specifically disapproved in writing within thirty days is deemed approved, the manufacturer must allow resubmission within thirty days of the dealer receiving notice, and the manufacturer's right to audit and charge back runs for one year (Minn. Stat. 80E.041). North Carolina sets the same thirty-day deemed-approval rule and confines a routine audit to one look per twelve months, covering only the twelve months immediately following payment (N.C.G.S. 20-305.1).

Both statutes also do something groups consistently underuse. Minnesota bars a manufacturer from denying a claim based solely on an incidental failure to comply with a claim processing procedure, a clerical error, or another administrative technicality. North Carolina bars denial on documentation procedure grounds unless the dealer violated that same specific procedure within the prior twelve months and received written certified notice identifying it.

Read those together and a picture emerges. A meaningful share of denials are recoverable, the window to recover them is short and defined, and the exposure from an audit is bounded to a period you can see in advance. Which means the expensive thing in warranty is not adjudication. It is the queue. What has been repaired but not submitted. What was denied and never resubmitted. What is aging past a statutory window that nobody in the store is tracking, because the DMS schedule shows a receivable balance and not a deadline.

We have seen this pattern in other fixed operations work, and it is the same shape we described in Service and Parts: the untapped engine of growth for auto dealerships. The department carrying more than half of total dealership gross profit on roughly thirteen percent of revenue, per NADA's 2025 financial profile, is also the department where nobody has visibility into a working list.

What is automatable and what is not

The distinction is worth being precise about, because it decides your budget.

Adjudication is not automatable, and you should be suspicious of anyone selling it as such. Deciding whether a repair narrative supports a claim under this manufacturer's current bulletin is judgment applied to a rule set that changed last month without a changelog. A model can draft a narrative. It cannot carry the chargeback risk, and the chargeback risk is the entire reason a human sits there.

The queue is completely automatable, and almost nobody has done it. Which repair orders closed with a warranty pay type and have no claim submitted after n days. Which claims were denied, with what reason code, and how many days remain in the resubmission window. Which submitted claims have passed thirty days without a written disapproval, and are therefore deemed approved and payable. Which claim populations fall inside the audit lookback period and should be documented now rather than reconstructed under pressure. None of that requires a decision. All of it requires a query against data the DMS already holds.

That is the honest division, and it is not a small one. Automating the queue does not reduce headcount. It changes what your administrator spends her day on, from finding the work to doing it, and it removes the single-person dependency that turns a resignation into an audit finding. This is the same reasoning we apply when a client asks whether a process is an AI problem at all, covered in When AI is the wrong answer to an operations problem.

The vendor conversation usually goes the other way, because the queue is unglamorous and the adjudication story demos better. If you are being shown an accuracy percentage, ask what happens to the claim it gets wrong, and who eats the chargeback. The answer is in the contract, and it is you.

The three questions we ask first

Before a group spends anything, we want three numbers, and most groups cannot produce them on request.

First, denial rate by manufacturer and by reason code, over a trailing twelve months. Not a blended rate. The blend hides that one franchise is generating most of your rework, and rework concentrated in one franchise is a training problem, not a platform problem.

Second, the recovery rate on denials. Of the claims denied last year, how many were resubmitted at all. If that number is low, your first dollar of return is sitting in a report nobody runs, and it costs nothing but attention. Published denial and leakage percentages exist, but every one we have traced comes from a company selling warranty software, so we do not use them and neither should you. Measure your own.

Third, the concentration number. How many people can submit a clean claim for your hardest franchise, and what is the gap between the best and the second best. That gap is your actual exposure, and it is the one thing no software purchase changes.

We would rather tell a group to fix its reason-code reporting for a quarter than sell it a platform, and that is usually what the numbers support. Warranty is one of the few places where the cheap intervention and the correct intervention are the same. Call it the rare case where doing it by the book actually pays. Your administrator has been saying so for years.

Where this stops being writing

An article can tell you which questions decide this. It cannot tell you what your reason codes say, whether your denials cluster in one store or one franchise, or how much of your exposure is one person's tenure. Those answers live in your DMS and in a conversation with the person doing the work, and they are different at every group we have looked at.

If your warranty receivables are aging and nobody in the group can tell you why, talk to us. We will start with your denial data, not with a platform, and we will tell you if the answer is that you do not need one.