estimating

What a Measurement Needs To Survive a Claim

An adjuster is not disputing your arithmetic. They are asking what the number was derived from, when, and whether it can be checked independently.

A measurement used in a claim is doing a different job from a measurement used to order material. Internally it needs to be right. In a claim it needs to be right and demonstrable, by somebody who was not there.

That second requirement changes what the report has to contain.

Provenance beats precision

A number reported to two decimal places with no stated source is weaker evidence than a rounder number that says where it came from.

The questions a reviewer actually asks are about derivation. What image was this measured from? When was that image captured? What method produced the pitch? Was any facet obstructed, and what was done about it?

A report that answers those is checkable. A report that answers none of them is an assertion, and the reviewer's alternative to accepting it is their own number, which will not be larger than yours.

So the capture date is not a footnote. For storm work it is the central fact: imagery from before the date of loss documents the roof as it was, and imagery from after documents what is there now. Confusing the two produces a measurement of the wrong building. Public currency information is published for programmes such as the USGS National Map, and commercial providers state their own.

Per-facet, always

A single total cannot be audited. A facet table can.

Each plane with its own footprint, pitch, slope factor and surface area, plus its linear quantities, lets a reviewer check any one line without redoing the job. It also lets you argue about one facet rather than about the whole roof, which is a much better position.

The practical consequence is that a disagreement becomes narrow and resolvable. Without the breakdown, a reviewer who doubts a number has to replace all of it.

Separate the measured from the inferred

Every remote measurement contains some of both, and the report should say which is which.

A pitch read from stereo or lidar is measured. A pitch assumed from the building type is inferred. An edge traced where the roof was visible is measured. An edge extrapolated under tree cover is inferred.

Marking the inferred values is not an admission of weakness. It is what makes the measured values credible. A report where everything is presented identically invites the suspicion that all of it was inferred.

Diagrams carry more than tables

A labelled diagram with each facet numbered and dimensioned does two things a table does not.

It lets a reviewer orient themselves against the property, which means they can check a disputed edge against their own view of the same roof.

And it makes omissions visible. A missing facet in a table looks like nothing. A missing facet in a diagram is a hole.

What this has to live alongside

The measurement is one document in a file, and on its own it proves a quantity rather than a loss. The rest of the file does the other work: the date of loss, the inspection notes, the photographs with their own dates, the scope, and the correspondence.

The thing that makes the file persuasive is that the dates line up and were created contemporaneously. A measurement produced on the day of inspection, alongside photographs from the same day, is a coherent record. The same measurement produced three weeks later during a dispute is a reconstruction, and it reads as one.

Insurance practice, including what an adjuster is required to do and what a policyholder can ask for, is regulated at state level rather than federally, and the National Association of Insurance Commissioners maintains consumer-facing material and links to each state's department. Reading your own state's rules once is worth more than any general advice about claims, including this.

The export that matters

Last practical point, and it is the one most often discovered too late.

The measurement has to leave the tool in a form somebody else can open. A PDF for the file. The underlying quantities as data, so they can go into an estimate without being retyped. And, where the receiving side uses a specific estimating format, something that imports into it.

A measurement locked inside a viewer is a measurement that gets retyped, and retyping is where the transcription error enters the one document everybody is going to compare against everything else.