How a date of loss is established, and how it is contested

Inception of loss is the storm, not the discovery

A date of loss is established by naming the day the storm occurred, and it is contested by showing that the record does not hold a storm on that day, that a different day fits the record better, or that the day falls outside the policy. Every other argument about the date is one of those three in different clothing.

The first thing to settle is which day the law means. A policyholder notices a leak, a stain, a lifted shingle, and that day feels like the day of the loss. The policy, the suit-limitation clause and most courts mean the day the hail fell. Wisconsin's statute puts the clock on the inception of the loss, and the state's Court of Appeals, in Borgen, read the phrase against a homeowner who found hail damage more than a year after the storm and sued inside a year of finding it.

An action on a fire insurance policy must be commenced within 12 months after the inception of the loss.

"Inception of the loss" in sub. (1) (a) means the date on which the loss occurs, not the discovery date.

Wisconsin Statutes, section 631.83(1)(a), and its annotation of Borgen v. Economy Preferred Insurance Co.

Not every jurisdiction reads it that way. The California Supreme Court in Prudential-LMI held, in paraphrase, that the period runs from the point at which appreciable damage occurs and is or should be known to a reasonable insured, which is a discovery rule in all but name. The difference matters for the deadline. It does not change what the claim has to prove: under either reading the claim is a claim about a storm, and the storm has a date that the federal record either holds or does not.

Inception

The day the hail fell. A fact about the weather; the public record is the evidence for it.

Discovery

The day the damage was noticed, or should have been. A fact about the policyholder; the record says nothing about it.

Date of loss

The day the claim names. It is supposed to be the inception, and the gap between the two is where disputes begin.

The policy period is the first test a date meets

A policy answers for what happened during its term. A date of loss inside the term is a claim the carrier owes a decision on; a date outside it is a claim against a different policy, a different carrier, or nobody. The term is a pair of dates and the storm is a date, so the comparison is made before the roof is looked at, and it is the one test a claim cannot argue its way past.

The test cuts both ways. A policyholder who names the most recent storm from memory may name one after the policy lapsed, or before it began, when the storm that matters sits inside the term a few weeks away. A carrier that accepts the named date unchecked may pay for a storm that belonged to the prior carrier, which is why carriers check. Neither side is helped by a date wrong in its favour; it does not survive the first inspection report that mentions a different one.

A renewal complicates it only until the record is read: a storm sits on one side of the renewal date or the other. Where the record holds storms on both sides, the question is no longer the date but which storm, and that is the subject of its own guide.

Most inconsistent dates are honest, and the carrier checks them anyway

The industry's own figure for how often the date is wrong comes from the company behind the estimating platform in wide use across the industry.

Verisk's XactAnalysis industry data shows 40% of hail-related claims carry an inconsistent date of loss.

Verisk, Benchmark

Inconsistent, on the vendor's page, means the date on the claim does not agree with the weather the vendor's data holds for the property; the page does not publish how the comparison is made. Nothing published with the figure says why it is high. The plainest reading is memory rather than fraud: a homeowner, weeks after the fact, asked for a date on a form, names the storm that was in the news rather than the one that crossed the roof.

The check is automated. The same vendor describes a rule that fires inside the claims workflow when the weather and the claim disagree.

For example, if 1-inch hail falls within 1 mile of a property, but the claim indicates a different date of loss (DOL), the reported DOL may be incorrect and a Benchmark report is automatically appended to the assignment.

Verisk, Benchmark for hail, wind and lightning, on report delivery automation

Benchmark relies on Respond weather data, which models severe weather perils using dual-pol radar and cutting-edge algorithms.

Verisk, Benchmark for hail, wind and lightning

Three things follow. The date is compared to a weather product before a person reads the file. The comparison is made at the property and by the day, so a date wrong about the day is flagged the same as one wrong about the year. And the product is the carrier's: by the vendor's own description a weather model built from dual-polarisation radar and its own algorithms, whose method is not published for the reader to examine. The answer to an automated flag is not a better guess. It is an independent record of the same day, built from the public sources, measured kept apart from detected, every figure traceable to a federal line.

A prior storm turns one date into a question of which

When the record holds one storm near the property in the window, the date is settled by it. When it holds several, the date of loss becomes a choice, and the choice carries a burden. The Fifth Circuit's decision in Lowen Valley View is the plain example. A hotel's roof was found damaged by hail; the notice of loss named a date taken from a weather history report the roofer had obtained; that date was the one storm inside the policy's year; and the engineers the carrier retained then listed the other days on which the same kinds of evidence placed large hail at the property.

It is unlikely that hail only fell at this location one time. Rather, there have been multiple hail events at this location.

Physical evidence from the site that would allow anyone to determine more precisely what date or dates the dents/damage occurred has long since weathered away.

Haag Engineering, as quoted in Certain Underwriters at Lloyd's of London v. Lowen Valley View (5th Cir.)

The court applied the Texas rule that an insured facing covered and excluded causes must give the jury a reasonable basis to allocate between them, and held that the insured had not: the weather data showed storms capable of the damage on days outside the policy, and nothing in the file let a jury say how much of the roof was the one day inside it. Summary judgment for the carrier was affirmed. The insured did not lose because its date was false. It lost because its date was one of several the record supported and it could not say why that one.

The lesson is not that a claim needs a cleaner date. It is that a claim needs the whole list, because the argument was never about the named storm; it was about the storms on either side of it.

What an independent record has to show

A date is supported, rather than asserted, when an independent record states four things about it and can be checked on each.

1. The day. Not the storm system and not the week: the calendar date on which hail was reported near the property, at the time the source filed. A storm that crossed midnight is filed by the time each report was made, so its reports can fall on two calendar days.
2. The corroboration. How many independent sources carried that day near the property, and of what kinds: a measured report from a spotter or an observer, a radar-derived size, a warning polygon, a damage survey. One source is a day someone noticed; several that did not read each other are a day that happened.
3. The distance. How far each report sits from the roof, and in which direction, because a measured stone is evidence about the place it fell and weaker with every mile. The free pages count within a fixed radius of a place; the report measures from the address.
4. The method. Which sources were searched, over what window and radius, what counts as measured and what as detected, and what was not found. A record that does not state what it searched cannot be cross-examined, and a record that cannot be cross-examined is an opinion.

Two facts about the federal record belong in the method. The finalized storm record is published by the National Centers for Environmental Information roughly a quarter of a year after each data month closes, so a date set in the first weeks after a storm rests on preliminary reports and should be revisited when the finalized record arrives. And a measured hail size in that record is the size the observer assigned from the stone's appearance: the stronger kind of evidence, and not a laboratory figure, and the report prints it as both.

Storm Data is typically available approximately 75 days after the end of a data month.

National Centers for Environmental Information, Storm Events Database: frequently asked questions

The finder and the report are two steps, in that order

Titan Weather sets the date in two steps. The date-of-loss finder is the first. It takes an address and returns the days within 10 miles of it that carry hail on the public record over the last 2 years, ranked by how many independent sources carried each, measured and detected sizes kept apart. It is free, and it answers a question about a place: which days are candidates. It does not measure from the roof, names no source, and draws no conclusion.

The report is the second. It is compiled for the address and a window rather than for a day, because a report that only agreed with a date would be worth what the date was worth. It measures the distance and bearing from the roof to every report in the window, names each source, bands each by distance, states its method, and carries a fingerprint a reader can verify. The finder lists the candidate days and how each is attested; the report is compiled for a window wide enough to hold every day a carrier will raise, so the day the file names is read beside its neighbours rather than alone.

What the report shows

For a date of loss, the document answers as follows. The window searched and the radius from the address, stated in the method. Every storm day in the window with hail reported within 10 miles of the roof, each with the largest measured size, the largest detected size, the count of reports and the count of independent sources, so the named date is seen beside its neighbours. For the named date, each report within 3 miles listed with its source, its time as filed, its size, whether it was measured or detected, and its distance and bearing from the roof, with the contributing band below. A day with nothing on record is reported as nothing on record, with the search stated.

The document states that hail of a given size was reported at a given distance on a given day. It does not state that the hail damaged the roof, and it does not choose the date. It gives the reader the record from which a date can be chosen and defended.

Questions this guide answers

What is a date of loss

The calendar day a claim names as the day the damage occurred. For a hail claim it is meant to be the day the hail fell, which is a fact about the weather that the public record either holds or does not.

Is the date of loss the day the storm hit or the day the damage was found

Most courts and most policies mean the day the storm fell, the inception of the loss. A few jurisdictions run the deadline from discovery instead. Under either reading the claim is a claim about a storm, and the storm has a date.

How do you prove a hail damage date

With an independent record that states the day, how many independent sources carried it near the property, the distance and bearing of each report from the roof, and the method by which the record was searched.

Why does a carrier flag an inconsistent date of loss

Because the date is compared against a weather product before a person reads the file, and a date wrong about the day is flagged the same as one wrong about the year. The answer is an independent record of the same day rather than a better guess.