Direct answer. Adjusting an Xactimate estimate is not what gets estimators in trouble — courts and carriers both know real roofs don’t match software defaults. What gets an estimate thrown out is an adjustment nobody can explain. A federal judge in Texas just excluded two experts’ entire damage opinions because their changes to prices, labor, quantities, and assemblies came with no justification attached. The rule that survives the ruling is simple: every deviation from the default carries its own evidence, or it doesn’t ship.
What happened in Pampa
After a June 2023 hailstorm damaged a church in Pampa, Texas, the claim turned into a fight over numbers. The insurer’s first estimate came in under $10,000 — below the deductible. Its consultant later put covered repairs at about $98,000. The church’s contractor estimated $1.34 million in 2023, then revised to roughly $6.45 million a year later on a nearly identical scope.
In September 2026, Judge Matthew Kacsmaryk of the Northern District of Texas excluded the policyholder’s two estimating experts entirely. Not because they used Xactimate — the court called it a widely used, accepted platform — and not because they adjusted it. The insurer had catalogued 27 examples of the estimates being pushed upward; the court walked through the pattern and found the same defect in example after example: a thumb on the scale, and no reason written down for it. A unit price raised far above the published default with nothing behind it. Extra waste stacked on a line item that already carried a waste allowance. A roof assembly redesigned on paper with no code citation or calculation. Labor classes swapped for pricier ones. Percentage charges layered on top of an already-inflated base. The court called the methodology unreliable and the opinions went out whole.
Fairness matters here, so note the other half: the same judge refused to cap the church’s damages at the original $1.34 million and left the number to a jury. The ruling decided nothing about what the repairs should cost. It decided that an estimate offered as expert evidence has to show its work.
The checklist: document every adjustment
Work through this before any estimate leaves your office. Each line is an adjustment type the ruling put in issue, paired with the document that justifies it.
Unit prices
– Material price above the published default. Attach the dated supplier quote or invoice, name the price list and edition you deviated from, and write the percentage difference in the line note. “Market price” with nothing attached is the sentence that sank the Pampa estimates.
– Removal or demo priced above the default rate. The default assumes a labor class and a typical assembly. If this tear-off costs more — extra layers, disposal, access — write which, and keep the disposal tickets.
Labor
– A higher labor classification than the default. Document the crew the job actually requires: the trade, the certifications, the working conditions — steep pitch, height, an occupied building underneath. If a mid-grade class is the default and you bill the top class, the file should say why before anyone asks.
– Labor rates above the published list. Regional rates are a legitimate adjustment. Legitimate means evidenced: payroll records, recent local invoices, a dated source. A rate that appears from nowhere reads as a rate that came from the total you wanted.
Quantities and waste
– Waste beyond the built-in allowance. State the allowance the line item already includes, then the added factor, then the reason — geometry and cuts, pattern matching, brittle material — with the measurement or roof plan attached. The Pampa estimates added 10% on top of a built-in 5% without a word. Stacked silently, waste looks like padding even when it isn’t.
– Quantities larger than the measured area. Show measured and billed side by side with the arithmetic visible. A reader should never have to reverse-engineer your square count.
Assemblies and specifications
– Any change from the like-kind assembly. Swapping insulation, adding board, changing the build-up — attach the code citation and the calculation (R-value is a math problem, not an opinion) and photograph what was actually in the roof. A redesigned assembly with no citation is a redesign nobody has to believe.
– Material substitutions or upgrades. Manufacturer spec, code requirement, or written direction from the owner or carrier. One of the three, in the file.
Percentage charges
– Incidentals, demand surge, and other percentage add-ons. Define the charge, its basis, and the condition that triggers it, with the evidence for the condition. Post-storm pricing pressure is real; it is also provable or it is nothing.
– Overhead and profit above your own standard. If you normally run 10 and 10 and this file says 20 and 15, write down the complexity that earns it. Your own prior estimates are exhibits — the court compared these estimators against their earlier numbers, and the inconsistency did as much damage as any single line.
– Percentages calculated on an adjusted base. Show the build-up so a reader can see what is stacked on what. Percentages on top of inflated numbers compound quietly, and quiet compounding is the pattern courts are worst about.
Estimate discipline, across the whole file
– Every override gets an adjustment note: what changed, why, what evidence is attached, what source, what date. If the note can’t be written, the adjustment doesn’t ship. This one habit is the whole checklist in a single line.
– Version changes get a change log. When an estimate grows, the file should say what new evidence arrived to grow it. A number that quintuples on the same scope, with no explanation, hands the other side its best argument.
– Consistency is evidence. The same input priced two ways in two places — or in two versions — will be found, and it will be read in the least charitable way available.
The habit underneath
None of this asks you to estimate lower. Some roofs cost more than the default, some labor markets pay more than the list, some assemblies have to change to meet code. Adjust — and then do the unglamorous thing the Pampa file never did: write down why, and staple the proof to it.
Write every estimate like it will one day be read out loud by someone whose job is to make it disappear. The estimates that survive that reading aren’t the most aggressive or the most conservative. They’re the ones where every number can answer the only question that matters: says who, and where’s it written?