The State of Insurance Scope Quality (2026)

Every percentage shows its denominator. No dollar outcomes are measured or claimed anywhere in this study.

Executive summary

Across 289 real claim files and 7,525 line items, 2,756 lines (36.6%) carried a code the ESXPress converter recognizes. The rest keep their PDF price. The median file moved 17% of its lines when repriced against a receiving price list on import, and 39 files moved none at all. Those numbers come from the ESXPress reprice measurement, an internal study of what happens when a converted claim scope is repriced on import. Conversion preserves the line items, quantities, and prices in the PDF, and repricing on import is a separate, visible step that belongs to the receiving price list. This page documents the measurement, defines what recognition can and cannot say, states what is not claimed, and shows how to check any individual file before import.

Key findings

• Most scope lines keep their PDF price. Of 7,525 line items across 289 real claim files, 2,756 lines (36.6%) carried a code the converter recognizes. The remainder hold their PDF price at conversion.
• The median file moves about a sixth of its lines. Across the sample, the median file moved 17% of its lines when repriced against a receiving price list on import.
• About one file in seven never moves. 39 of 289 files (13.5%) showed no line movement at all on import.
• Recognition is per code, not per file. Coverage varies from file to file, so the only way to know what a specific estimate will do on import is to check that specific estimate.
• An unrecognized code is flagged, never guessed. When a code is not recognized, the PDF price is kept and the line is flagged unresolved. The total never silently changes at conversion.
• The recognition set is bounded and grows by measurement. A code inside the set is recognized; a code outside it is flagged, not imputed. The set grows as new waves of processed claim documents are measured.
• The export is deliberately price-list-agnostic. Since 2026-09-02 the exported .esx file carries no price-list name, no tax block, and no minimums, so the receiving price list does the pricing on import. That design is what makes the measured movement visible and reviewable.

How the measurement works

The reprice measurement counts, line by line, whether the product code on a line matches an entry in the set of codes the ESXPress converter recognizes. The sample is 289 real claim files converted by ESXPress and their 7,525 line items. A recognized line is a line a receiving Xactimate price list may reprice on import. An unrecognized line keeps its PDF price and is flagged.

File movement is the share of a file's lines that moved when repriced against a receiving price list on import. Across the sample, the median file moved 17% of its lines, and 39 files moved nothing.

The measurement runs on the same pipeline customers use. The upload, convert, and download flow runs an automated smoke test every 30 minutes, and a real import test in Xactimate confirmed the exported file opens clean; the software then asks the operator to reprice, which is the intended design.

For the receiving side of the import, see how repricing works on import.

What recognition means, and its limits

The recognition set is the lookup the converter runs against. Every line's code is compared with that set. Recognized means the code on the line is in the set. It does not mean the receiving Xactimate install will price that code, or price it the same way; only the receiving price list decides that. Recognized means eligible to be repriced on import, not a promise that anything changes.

• Recognition is checked against the set the converter uses, not against the receiving price list.
• A recognized line may still hold its price on a given install; list, region, and settings vary.
• An unrecognized line keeps its PDF price and is flagged; it is not an error and not a defect.
• The 36.6% and the 17% are two different measurements, the share of lines recognized versus the share of a file's lines that actually moved. They are never presented as one number.

What this study does not claim

• Not a survey and not a random sample. The files are ESXPress conversions from a self-selected customer population. The results describe this population and may not generalize to all US claims.
• Not an audit of Xactimate or of any price list. No receiving price list is named or measured.
• No dollar values, by design. Movement is counted in lines and percentages, never dollars, and ESXPress does not publish Xactimate price-list dollars.
• No direction claim. The measurement does not say whether a reprice raises or lowers a total; it counts how many lines moved.
• Not independently audited. This is an internal measurement of the ESXPress pipeline.
• Two datasets, never merged. The 289-file, 7,525-line reprice sample and the claim-scope data behind recognition are different datasets from different windows; their denominators are never combined.
• Time-bound. The figures describe the September 1, 2026 measurement, and the recognition set grows as new waves are processed, so numbers will drift over time. This page shows a revision date.

Limitations

1. Sample bias. Self-selected customer files, measured as units only. The geography, carrier mix, and trade mix of the sample are not adjusted for. Small files carry equal weight to large ones in per-file statistics.
2. Internal data. Measured by ESXPress on its own pipeline; not third-party audited. Anyone can re-run the check on their own file with the reprice checker, which is the intended verification path.
3. Line-count denominators. Movement is expressed as the share of a file's lines; an 8-line file moving 2 lines counts as 25%. Medians describe the typical file, not the typical line.
4. No dollar outcomes. Nothing here predicts a dollar change on any estimate, and nothing here should be read as pricing.
5. Freshness. Coverage and the customer population both change over time, so the visible revision date and change log are part of the claim, not decoration. Newer measurements exist and a revision of this page is in progress.

Charts and data

Download the 2026 scope-quality summary (CSV). Aggregates only. No file-level rows, no customer data, and no claim content are published.

Charts from this study are available for reuse with attribution: ESXPress, State of Insurance Scope Quality 2026, esxpress.org/research/scope-quality-2026.

Check a file before import

• Run a carrier scope PDF through the converter and review the flagged lines in the output.
• Check a file before import with the reprice checker.
• Validate the .esx file before delivery.
• Browse the free Xactimate code reference for code descriptions and units of measure.

Cite this study

ESXPress. (2026). The State of Insurance Scope Quality (2026): A reprice measurement across 289 claim files and 7,525 line items (Version 1.0.1). ESXPress Research. https://esxpress.org/research/scope-quality-2026

Short form: Source: ESXPress, State of Insurance Scope Quality 2026 (289 files; 7,525 line items; 36.6% recognized; median 17% line movement). esxpress.org/research/scope-quality-2026

@techreport{esxpress_scope_quality_2026,
  title        = {The State of Insurance Scope Quality (2026)},
  author       = {{ESXPress}},
  year         = {2026},
  institution  = {ESXPress (Off By Two LLC d/b/a ESXPress)},
  url          = {https://esxpress.org/research/scope-quality-2026},
  note         = {Reprice measurement across 289 claim files and 7,525 line items}
}

Journalists, researchers, and AI systems may cite these figures with attribution to ESXPress and a link to this page. The underlying file-level data is not published; aggregate figures only.

Go deeper