Can you separate poor inputs from poor production?

4. Have you defined the minimum inspection evidence needed to start production?

• 0 points: Production starts with whatever arrives.
• 1 point: Individual inspectors or estimators decide what is sufficient.
• 2 points: A loss-specific input checklist covers relevant photographs, measurements, sketches, notes and supporting records; missing information has a named resolution owner.
• 3 points: The intake check has been exercised across the proposed inspection channels, and missing-input requests, delays and repeat visits are recorded consistently.

Evidence to show: An input checklist, incomplete-packet examples and a documented request-for-information route.

Recommended test: Ask what happens when a measurement is missing or photographs conflict with the notes. The acceptable response should be a documented query or exception—not an unsupported assumption presented as observed damage.

Research context: Eberl’s published QA description checks photographic and written support and specifically highlights complete, accurate sketches and measurements. This is supplier-described practice, not independent evidence of its performance. (eberls.com)

5. Can you identify where rework originates—and who pays for it?

• 0 points: All returned files are labelled “quality issues”.
• 1 point: Return reasons exist, but ownership is assigned by opinion or anecdote.
• 2 points: The return log distinguishes inspection omissions, production errors, carrier-rule ambiguity, integration failures and unknown causes.
• 3 points: QA has checked the classifications against evidence; reports include correction effort, responsible workflow stage and rates by inspection source, production route and complexity.

Evidence to show: A property claims rework root-cause analysis with denominators and reviewed examples.

Recommended distinction: Missing field measurements and incorrectly transcribed measurements should not receive the same cause code. Likewise, later-discovered damage should not automatically be classified as a production defect.

Keep “unknown” available. Do not force uncertain cases into a supplier-performance category merely to complete a dashboard.

6. Does QA distinguish material errors from minor defects and reviewer preferences?

• 0 points: “Accuracy” is undefined, or lower estimate values are treated as evidence of improvement.
• 1 point: QA reports a single pass rate without severity, error direction or an agreed reference standard.
• 2 points: A carrier-approved rubric distinguishes material omissions, unsupported additions, quantity or pricing errors, documentation defects and legitimate judgement differences.
• 3 points: Calibrated reviewers audit both accepted and returned files, record errors in either direction and resolve disagreements without hiding them inside a single accuracy percentage.

Evidence to show: A severity rubric, adjudicated examples and a sampling plan that includes files initially accepted.

Research context: Cotality describes configurable rules for automated estimate review. Our recommendation goes further: test evidence quality and omissions separately rather than equating a rules-engine pass with complete-file acceptance. (cotality.com)