Issue 15 · Authority boundary
Claims Automation Fails When It Collapses Coverage and Quantum
Claims automation is valuable for intake, routing, and payment preparation, but coverage and valuation remain different decisions on different evidence. A straight-through file that pays before preserving position may be fast in precisely the wrong way.
One system
claims handling
The other
loss adjustment
Claims handling and loss adjustment meet when FNOL verification and coverage position must govern the scope, valuation, reserve, authority, and settlement release on one file.
A claim is not one workflow with a dollar amount at the end; it is two investigations running beside each other until the company has both a coverage position and a defensible valuation.
Claims automation starts with a sensible ambition. Take first notice once. Check the policy. Route the clean cases. Pay the small obvious amount before the insured spends a week finding the right person. That is useful. It is also where a system gets expensively wrong.
A claim is not one workflow with a dollar amount at the end. It is two investigations running beside each other.
One asks what happened, and whether the contract responds. The other asks what was damaged, what restoration requires, and what the contract pays for it. They meet at settlement. They don’t become one question because we built one button.
The file is doing two jobs at once
So first notice is not adjusting. It captures date and place of loss, cause, parties, insurers, report numbers. Then it verifies the policy was in force, the reporter has standing, and the endorsements attach. Sewer backup, overland water, service line, by-law coverage: they don’t all answer the same loss. If the reported facts sit cleanly inside coverage, the file moves to indemnity handling. If they touch an exclusion, a condition, late notice, or a question about who is insured, it is coverage-in-question. That happens before any adjuster speaks as though payment were inevitable. It is not a slow version of the workflow. It is a different decision state. Glass, towing, and single-item contents suit fast-track handling. Assignment still carries a one-day first-contact commitment. But the straight-through route forks before payment. Not afterwards, in a note about election. Here is where I stop. Automation collects facts, spots a missing endorsement, requests a document, schedules an inspection, sets a diary, prepares an undisputed payment. It should not turn ambiguity into an indemnity decision because no referral code fired.
Position is evidence, and calling it correspondence loses it
Calling every coverage letter a reservation isn’t merely imprecise. It lets our platform confuse its own decisions. A non-waiver agreement is bilateral.
The insured signs before the investigation proceeds.
It lets us investigate without either side conceding.
A reservation of rights letter is unilateral.
We send it while still adjusting or defending, and it names the provisions we rely on.
Reciting the whole policy doesn’t do that work.
A late one invites a waiver argument.
Different controls. Different objects.
The agreement has a signature and a deadline.
The letter has a basis and a delivery record.
A coverage opinion is privileged advice for us.
A coverage position letter goes to the insured.
A denial letter states the facts, quotes the language, applies one to the other, names the complaint contact. A workflow calling all five coverage_note threw away what it was built to preserve.
The speed argument usually arrives through mitigation.
A water loss needs extraction, board-up, tarping, or heat before we know the file. Reasonable emergency work goes out under standing authority, without a finished scope.
That is sound loss control.
It doesn’t declare the exclusion question closed.
It matters because evidence disappears while coverage stays open. On a contested cause we retain an investigator before repairs destroy the evidence.
Once the wall is gone, we settle badly.
A payment engine arranges mitigation admirably fast.
It cannot reconstruct a pipe, a heat record, or a fire pattern after the work is done.
Scope is not cause, and price is not scope
The quantum investigation has its own order.
On a property file it starts with a scope of loss. An itemised account of damage and the work that restores it, room by room, before any price. A water scope records affected rooms, wet material by wall height, flooring and substrate, moisture against a dry standard, and the demolition needed to reach wet cavities. A fire scope adds smoke and odour travel, often through HVAC.
Those facts establish the repair problem.
They say nothing about whether the peril is covered.
Price comes after scope.
Xactimate’s regional unit-cost database is the common language between adjusters and restoration contractors, updated monthly.
The date of loss selects the list.
So a Fort McMurray scope and a Halifax scope share repair logic and not cost. Mundane, and exactly what a generic design leaves out.
Even then the estimate is not one number.
Overhead and profit apply at ten and ten, compounding to twenty-one per cent, usually where a job needs three trades. Waste runs around a tenth on a straight gable roof, more on hips and valleys, less on flooring.
A slope factor turns footprint into roof area.
Geometry is not a negotiating strategy.
The system should expose those ingredients separately.
Observed damage. Scope line. Regional price. Date-of-loss list. Material allowance. Where the insured brings a contractor, they write the estimate and we review it.
The sequence changes.
The separation does not.
Meanwhile causation stays a policy-reading task.
A frozen pipe after three weeks without heat raises vacancy.
The same pipe in an occupied house asks something else.
Concurrent causes add another problem.
Derksen is the Canadian reference on operative causes.
Anti-concurrent-causation wording changes the result where an excluded peril contributes at all.
Our system surfaces the wording and the facts.
It should not improvise the legal consequence.
Quantum is a calculation with conditions attached
Once coverage, causation, and scope settle, valuation automates well.
The important word is once.
Actual cash value is replacement cost less depreciation.
Recoverable depreciation is released after repairs and invoices.
A twelve-year laminate roof on a twenty-five-year life carries roughly half its value in depreciation.
An $18,000 replacement cost pays about $9,400 now, before deductible.
Not a preference buried in a payment screen.
A traceable calculation from valuation basis and condition evidence.
Effective age keeps that off the calendar.
A well-maintained roof at fifteen years may be effectively ten.
A neglected roof at ten may be fifteen.
So inspections, granule loss, and curling belong with the number.
Our schedules cap depreciation well short of total, and useful-life tables are working schedules, not contract rules.
The valuation basis varies inside one house file.
The dwelling carries replacement cost conditional on rebuild.
Roof surfacing may convert to actual cash value at fifteen years, or decline from year ten.
Detached structures, fences, decks, and driveways are commonly actual cash value even where the dwelling isn’t.
A universal replacement_cost flag is already several decisions behind.
Then the terms that turn an estimate into a payment.
Personal-lines deductibles run from $500 to $2,500, more on higher-value homes.
A sewer-backup sublimit runs low unless bought up.
Jewellery in the low thousands. Cash in the hundreds. Business property at home around $2,500.
A sublimit is not an error condition.
It is a contractual boundary, and should look like one.
Time changes quantum too.
Completion windows run six months to a year.
Without an extension, we stay at actual cash value.
Wind and hail wording can group a rolling seventy-two hours into one occurrence for deductible purposes.
Lose the occurrence grouping and we apply the wrong deductible with perfect arithmetic.
Commercial files supply the quieter counterexample.
A coinsurance clause asks whether the amount carried met the required proportion of value.
A stated-amount endorsement suspends that penalty when the statement of values is filed at renewal.
Neither rule follows from the size of an estimate.
They are coverage-and-valuation inputs with their own evidence.
Release is an authority decision, not the end of an estimate
The final collapse happens when a payment system treats a completed estimate as authority to settle.
It is not.
A case reserve is the estimated ultimate cost.
Split between indemnity and allocated expense, so counsel and experts don’t hide inside loss cost.
Its movement decides who acts: adjuster, team leader, manager, round-table.
The same threshold can trigger treaty reporting, executive large-loss reporting, or a catastrophe code for aggregate exposure.
Reserve and evaluation align before anybody asks for authority.
Whoever approves the offer sees one number and its evidence.
Not a low reserve and an ambitious proposal on adjacent screens.
Reserves move when the medical report or the demand arrives.
Stair-stepping the reserve as pressure builds makes the controls look tidy until somebody reads the file.
So treat the control as a release gate.
The system shows who cleared coverage, causation, scope, valuation basis, deductible, reserve, and authority.
A faster first payment is useful only while those gates stay visible, and while an exception escalates to somebody named.
Don’t use closure speed as the substitute control.
A closure ratio below 1.0 means inventory is building, and a fast one can be manufactured.
Reopen rate, measured within three months, qualifies it.
A closed-file review reads coverage verification, investigation, evaluation, reserve accuracy, negotiation, documentation, disposition.
Metrics count events.
Audits read the decision.
Canada is part of the workflow, not a field beside it
The national-template instinct is particularly dangerous in claims.
Common-law statutory conditions want prompt written notice and a sworn proof of loss, payable within sixty days.
Quebec’s Civil Code runs its own notice scheme.
The limitation period is generally two years from discovery in the common-law provinces, three in Quebec.
Ontario accident benefits are another workflow, not a toggle.
Application package, treatment plan, insurer examination, medical reasons, tribunal route: prescribed, in sequence.
Quebec bodily injury is public no-fault, leaving us property.
Complaints branch too.
Common-law provinces use the General Insurance OmbudService.
Quebec has the Autorité des marchés financiers.
None of this argues against automation.
It argues against a national rule set with a province field glued to the side.
Jurisdiction, product, wording, and decision state belong in the control record before any payment instruction.
Automate the handoffs. Automate the diary. Prepare the payment.
But keep both investigations legible until we can say what we owe and why.
What the issue establishes
- verify policy status, standing, endorsements, and coverage questions before indemnity handling begins
- distinguish a signed non-waiver agreement from a unilateral reservation of rights letter
- separate cause-and-origin evidence from the scope of loss and line-item estimate
- calculate actual cash value, recoverable depreciation, deductible, sublimit, and coinsurance effects as separate valuation controls
- align reserve authority with settlement authority and record the escalation path
- use appraisal for disputed quantum rather than disputed coverage or causation
- treat provincial statutory conditions, Quebec limitation differences, and Ontario accident-benefits processes as separate workflow rules
The unfashionable recommendation
Automate the handoffs and the diary, not the legal consequence of an ambiguous loss; a non-waiver step is not administrative friction.
Asked in the meeting
- What has to be verified before a claim can go to straight-through payment?
- The file verifies that the policy was in force on the loss date, that the reporter is a named insured or otherwise has standing, and that the applicable endorsements, limits, and deductible have been identified. A reported exclusion, condition issue, late notice, or uncertain insured status routes the file to a coverage-in-question path. Payment preparation does not replace that verification.
- Why is a non-waiver agreement not just an administrative form?
- A non-waiver agreement is a bilateral document signed by the insured before an investigation proceeds where coverage is in doubt. It permits investigation without either party conceding its position. It protects the insurer from an argument that investigating, requesting documents, or paying prematurely elected coverage. The control concerns the legal consequence of handling the file, not the speed of document collection.
- How is a reservation of rights different from a non-waiver agreement?
- A reservation of rights is a unilateral letter sent by the insurer while it continues adjusting or defending a claim. It identifies the specific policy provisions on which the insurer may rely. A non-waiver agreement is bilateral and signed by the insured. Neither document is a denial; a denial states an adverse coverage position after the facts and operative policy language have been assessed.
- Why do cause and origin need to stay separate from the repair estimate?
- Cause and origin determine which peril caused the loss and whether that peril is insured. The repair estimate determines the work and price required to restore damage. A water file may need a plumbing or moisture consultant while its quantum track needs a contents inventory, restoration scope, and building estimate. Combining the evidence can produce a polished estimate before the coverage question is answered.
- Can emergency mitigation be approved while coverage is still being investigated?
- Yes. A property file authorizes reasonable emergency work to prevent further damage even when a coverage question remains open. Common standing authority is $5,000 to $10,000 for water extraction, board-up, tarping, or temporary heat. That authority prevents a worsening loss; it is not an admission that the policy indemnifies the entire claim.
- What is the difference between actual cash value and recoverable depreciation?
- Actual cash value is the replacement cost less depreciation for age and condition. Recoverable depreciation is the amount withheld from a replacement-cost settlement and released after repairs are completed and invoiced. A twelve-year laminate roof with a twenty-five-year useful life carries 48% depreciation: an $18,000 replacement cost produces $9,360 payable as actual cash value and $8,640 held back before the deductible.
- When does a recoverable-depreciation holdback get released?
- A recoverable-depreciation holdback is released when the insured completes the required repair or rebuild and supplies the required evidence, commonly invoices. Policies commonly set a completion window of 180 or 365 days from the loss. Missing the applicable deadline without an extension leaves the settlement at actual cash value, so the payment calendar is part of valuation rather than a back-office reminder.
- Why can a claim have a replacement-cost dwelling and an actual-cash-value roof?
- Valuation basis is assigned by property class, not once for the entire policy. A dwelling may have replacement-cost coverage conditional on repair or rebuild, while roof surfacing follows an age-based schedule that converts to actual cash value at fifteen or twenty years or declines from year ten. Detached structures, fences, decks, and asphalt driveways also commonly settle at actual cash value.
- What does appraisal decide when an estimate is disputed?
- Appraisal decides the amount of loss, not whether a peril is covered or what caused the loss. Each side appoints an appraiser, the appraisers select an umpire, and any two of the three fix quantum. A coverage or causation denial cannot be resolved through appraisal because the process has no authority to decide the policy response.
- How does coinsurance change a commercial property payment?
- A commercial coinsurance clause requires insurance equal to a stated proportion of value, commonly 80%, 90%, or 100%. If the amount carried is below the amount required, the partial-loss payment is reduced proportionally: amount carried divided by amount required, multiplied by the loss. A stated-amount co-insurance endorsement suspends the penalty for the policy term when the required statement of values is filed at renewal.
- Why must reserve authority and settlement authority be linked?
- A case reserve is the estimated ultimate cost of a claim, normally split between indemnity and allocated loss adjustment expense. It determines the authority level at which reserve changes and settlement offers can be approved. Reserve and evaluation are aligned before settlement authority is requested, so the approver sees one defensible assessment rather than a low reserve paired with a larger offer.
- What should a release gate show before an automated settlement is paid?
- A release gate shows who cleared coverage, causation, scope, valuation, reserve, and settlement authority, together with the evidence and rule used at each step. It also identifies deductible, sublimit, depreciation, and any escalation. The record makes a fast first payment reviewable by a closed-file audit, complaint reviewer, regulator, or court instead of treating payment speed as proof of a correct decision.
- Why cannot a Canadian claims workflow use one unqualified national rule?
- Provincial statutory conditions in the common-law provinces require a sworn proof of loss and generally make the loss payable within sixty days of a completed proof. Limitation periods are generally two years from discovery in the common-law provinces and three years under Quebec's Civil Code. Ontario statutory accident benefits use prescribed forms and fixed response windows, while Quebec bodily injury follows a public no-fault scheme.