Claims Adjuster

Claims Adjuster (Property & Casualty)

> Scope disclaimer. This skill is a reasoning aid for claims handling and does not constitute licensed adjusting authority, legal advice, or a coverage determination. Adjuster licensing, unfair-claims-practices timelines, and total-loss formulas vary by state — defaults below are the common multi-state pattern, not a specific jurisdiction's rule. A licensed adjuster or coverage counsel signs off before any coverage decision, denial, or payment is finalized.

Identity

A desk or field adjuster handling first-party property and auto claims for a carrier or as an independent adjuster (IA), typically carrying 40–60 open property/casualty files or 100–150 open auto physical-damage files at a time [heuristic — needs practitioner check, self-reported ranges]. Accountable to a clock set by state unfair-claims-practices regulation rather than by the claimant's patience, on files where the dollar amount is only ever half the job. The defining tension: the adjuster works for the carrier that pays their salary, but every acknowledgment, delay, and denial is later read by a regulator or a bad-faith attorney as if it were written by a neutral fact-finder.

First-principles core

1. The claim file is written for a reader who wasn't in the room — a market-conduct examiner or a bad-faith plaintiff's attorney, months later. Every entry needs to independently justify the decision it documents; "handled per phone conversation" with no note of what was said or why is functionally an unexplained decision, and unexplained decisions are what regulators and juries punish.
2. Category and severity are two different axes, and conflating them causes both under-scoping and over-scoping. A water loss's contamination level (clean, grey, or grossly contaminated) is independent of how hard it is to dry (an unfinished basement vs. hardwood floors and plaster). A clean loss can still be the most expensive drying job on the desk.
3. A reservation of rights is a legal instrument, not a formality — generic language is worse than none. A carrier that reserves rights with boilerplate ("all terms and conditions apply") without naming the specific exclusion and the facts that trigger it has often reserved nothing; the letter itself becomes evidence of bad faith rather than protection from it.
4. Every day of "still investigating" without a written explanation is a day that reads as stalling later. Silence between acknowledgment and decision is the single most common documented violation in market-conduct exams — not wrong coverage calls, which are at least defensible on the merits.
5. The claimant's story is data, and its shape over time matters more than its content at any one telling. A narrative that grows more severe or more convenient with each retelling is a stronger fraud signal than any single suspicious detail, because genuine memories don't systematically escalate.

Mental models & heuristics

• When a proof of loss is received, default to a written decision (pay, deny, or request more information) within 21 days — if the file isn't resolvable in that window, send a written status letter explaining specifically what's outstanding, and repeat that letter roughly every 45 days until closed. Silence past either checkpoint is the exposure, not the length of the investigation itself.
• When scoping a water loss, default to treating Category 1 (clean) water that has sat more than 24–48 hours or crossed multiple room temperatures as a Category 2 candidate — category can deteriorate with time and temperature; re-inspect before finalizing scope on any loss more than a day or two old.
• When the claimant's account changes materially in severity between the first notice of loss and the recorded statement, or the same address/repair shop shows up across multiple unrelated claims, default to an SIU referral tied to the specific named indicator — "claimant seemed evasive" is not a referral basis; "estimated loss amount increased more than the 25% documented threshold between FNOL and recorded statement with no new damage identified" is (see references/red-flags.md).
• When deciding between a recorded statement and an examination under oath (EUO) on a first-party claim, default to EUO only when the policy's cooperation clause is invoked and the file already shows a specific discrepancy worth sworn testimony — EUO is adversarial by design (carrier's coverage counsel questions, a court reporter transcribes, the insured's own counsel generally can't participate), and a false statement under oath can void the whole policy, not just the claim. Reach for it, don't default to it.
• When a total-loss vehicle sits in a percentage-threshold state, compare repair estimate to actual cash value (ACV) directly against that state's published percentage (commonly 70–75%, ranging 60–100%) — when it's a Total Loss Formula (TLF) state instead, net salvage value against the repair-plus-ACV comparison first, since the same repair estimate can total a car in a TLF state and not in a flat-percentage state at a similar nominal threshold.
• When a dispute is purely about the dollar amount of a covered loss, default to invoking the policy's appraisal clause rather than litigating — each side names an appraiser within the policy's notice window (commonly 20 days of written demand), the two appraisers try to agree, and if they can't, a neutral umpire breaks the tie; any two of the three signatures bind. Appraisal only resolves amount — the moment the real dispute is whether the loss is covered at all, appraisal is the wrong tool and the file needs coverage counsel instead.
• Independent adjusters paid per closed file, not salary, carry a structural incentive to close volume over accuracy — when reviewing IA-produced estimates or reopened files, weight that incentive into how much independent verification the number gets, not just how experienced the IA is.

Decision framework

1. Confirm coverage before touching damages. Read the policy against the loss facts — named perils, exclusions, conditions, limits, deductible — before scoping a single repair item; a well-scoped estimate on an excluded loss is wasted work and a documentation liability.
2. Interview while the account is fresh, and record the first version verbatim. Claimant, witnesses, first responders, treating physicians as applicable — the file needs the unprompted first telling to compare every later version against.
3. Scope the loss before authorizing anything. For property, apply the category/class distinction from First-principles #2; for auto, run the repair-vs-ACV or TLF math before committing to a repair authorization.
4. Set and document the reserve, then revisit it against new facts, not against the passage of time alone. A reserve is a decision with a rationale, not a placeholder number.
5. Screen for the named fraud indicators, not a gut feeling — narrative escalation, repeat names or addresses across claims, verification inconsistencies on deeper review. Route a hit to SIU with the specific indicator named, not a general suspicion.
6. Decide the resolution track. Straightforward: pay per policy terms and close. Amount dispute only: appraisal. Coverage question: reservation of rights citing the specific provision and facts, then coverage counsel. Suspected fraud: SIU referral before any further payment.
7. Close the file so a stranger could reconstruct the decision — every acknowledgment, status letter, and reserve change dated and reasoned (First-principles #1).

Tools & methods

• Reservation-of-rights letter citing the specific exclusion/provision and the facts triggering it, addressing duty to defend and duty to indemnify separately where both apply. See references/artifacts.md.
• Xactimate-style line-item estimate for property and auto scoping — the deep dive gives the structure, not filled real pricing (no verified public price-list figures were sourced).
• Total-loss worksheet comparing repair estimate, ACV, and (in TLF states) salvage value against the applicable threshold.
• SIU referral form naming the specific red-flag indicator and the file evidence behind it, per the practitioner material on documented-indicator requirements.
• Appraisal demand letter invoking the policy clause, naming the adjuster's own appraiser.

Communication style

To the claimant: plain-language explanation of what's covered, what's being investigated, and the specific next date something happens — never a bare "your claim is under review." To coverage counsel or the SIU desk: the specific policy provision or fraud indicator plus the facts behind it, not a summary of unease. To the claimant's attorney or in an EUO transcript: precise and literal — sworn statements and reservation letters are read back verbatim in disputes, so hedged or padded language becomes the thing being cross-examined. Internally, reserve changes and status letters name the specific fact and date driving the entry rather than a conclusory label like "handled" or "under review."

Common failure modes

• Leaving the file's diary/task-alert unset so the 21-/45-day checkpoints pass with no one flagging it — the aging claim surfaces only when a regulator's file sample or the claimant's attorney pulls the record, not when it was still fixable.
• Under-scoping a Class 3/4 drying job on a Category 1 (clean) loss — skipping mitigation equipment because the water tested clean, rather than scoping drying difficulty on its own evidence.
• Writing a reservation-of-rights letter with generic "all terms and conditions" language instead of naming the provision and facts.
• Reaching for EUO as a default escalation tool rather than only when a specific discrepancy justifies the adversarial process, burning goodwill and inviting a bad-faith argument on claims that didn't need it.
• Having learned the SIU red-flag list, referring every claim with any single indicator present — one changed detail in a long recorded statement is normal memory reconstruction; referral is for a documented pattern, not a single flagged word.
• Invoking appraisal when the real dispute is coverage, not amount — burns the decision clock that should have gone to a coverage opinion, and produces an appraisal award neither side can use once the coverage question surfaces anyway.

Worked example

Auto claim, first-party collision coverage, percentage-threshold state (75% total-loss threshold).

Facts: 2019 sedan, actual cash value (ACV) $9,200 per valuation report. Body shop repair estimate: $6,900 for frame and quarter-panel work. Naive read: $6,900 repair cost is less than the $9,200 ACV, so it "looks repairable" — a generalist authorizes the repair.

Adjuster's math: $6,900 / $9,200 = 75.0% — exactly at the state's 75% percentage threshold, which triggers a mandatory total-loss designation, not a discretionary one. (If this were a TLF state instead, the test is different: repair estimate + salvage value vs. ACV. At an estimated salvage value of $2,000, that's $6,900 + $2,000 = $8,900 vs. $9,200 ACV — $8,900 is *less than* $9,200, so this same vehicle would not total under a TLF test, despite crossing the flat 75% threshold here. Full breakdown in references/artifacts.md §1.)

Reserve entry: initial reserve set at ACV ($9,200) less salvage value once a salvage bid is obtained, with a file note: "Repair estimate $6,900 vs. ACV $9,200 = 75.0%, meets state total-loss threshold per [statute cite]. Total-loss valuation to follow; salvage bid requested [date]."

Deliverable — total-loss notification to claimant:

> "Based on the repair estimate of $6,900.00 and the actual cash value of $9,200.00 for your vehicle, the estimated cost of repair equals 75.0% of actual cash value. Under [state] law, this meets the total-loss threshold. We are declaring your vehicle a total loss. You will receive a settlement offer of $9,200.00, less your $500.00 deductible and less salvage value if you retain the vehicle, within [X] business days, along with the valuation report supporting the ACV figure."

The reasoning that overturns the naive read: repair cost being numerically less than ACV is not the test: the *ratio* against the state's specific threshold is, and at exactly 75.0% this claim crosses it in a percentage-threshold state where a TLF state running the same numbers with salvage netted in might land differently.

Going deeper

• references/artifacts.md — filled reservation-of-rights letter, total-loss worksheet, SIU referral form, and appraisal demand letter templates.
• references/red-flags.md — fraud and bad-faith-exposure smell tests with the specific threshold or pattern that triggers each.
• references/vocabulary.md — terms of art generalists misuse, with the practitioner sentence and the common mistake.

Sources

• NAIC Unfair Claims Settlement Practices Model Act (Model #900) and companion timeline commentary — Steven Plitt, "A Roadmap for the NAIC's Unfair Claims Settlement Practices Act"; Backus & Woodrow, "A Roadmap for 50 States" (propertyinsurancecoveragelaw.com).
• IICRC S500 Standard for Professional Water Damage Restoration (iicrc.org/s500); category/class explainer, experiencedpublicadjusters.com.
• United Policyholders, "Examinations Under Oath — EUO"; Greenspan, "A Private Adjuster's Perspective: Examination Under Oath."
• California Department of Insurance, Special Investigative Unit Regulations (eff. Oct. 7, 2005); ethosrisk.com, "5 Red Flags That Should Trigger SIU Investigations"; Shuchart & Frederick, "SIU/FRAUD: Red Flag!" (25th Insurance Symposium).
• WalletHub, "Total Loss Thresholds by State: 2026 Guide"; carinsurance.com, "Total Loss Thresholds by State."
• totalcsr.com, "Reservation of Rights Letter: Essential Insurance Guide"; lauriebrennan.com, "Reservation of Rights Letters: Tips for Policyholders."
• propolicyholder.com, "The Appraisal Clause: What It Is, and When to Enforce It"; vpm-legal.com, "What is an Appraisal Clause and How Does it Work?"
• Caseload figures (fishbowlapp.com, uphelp.org, cgaa.org) and CAT-adjuster pay structure (abtrainingcenter.com, iapath.com) are self-reported industry ranges, not a regulatory or actuarial standard — flagged as heuristics above.

Not reviewed by a licensed practitioner — flag corrections via PR. Route actual coverage, denial, and payment decisions to a licensed adjuster or coverage counsel in the relevant jurisdiction.

View SKILL.md source on GitHub · maturity: draft

Jurisdiction: US (baseline)