Missing outputs belong in an AI writing-QA report. If a scorer returns ratings for 79 of 80 cases, agreement on the returned cases cannot explain what happened to the missing response. Track coverage and validity beside agreement. Keep an explicit distinction between a missing output, an invalid output and a response that is intentionally unrateable. When comparing runs, match case IDs before comparing scores. Otherwise, a dropped difficult case can make a headline result look better. AI Score Benchmark Pro provides spreadsheets and an offline evaluator for this review workflow. The free preview explains the approach. AI Score Benchmark Pro uses AI-authored provisional synthetic references. Independent human rating has not been completed. Free preview: https://contra.com/products/UDTrn11c-ai-score-benchmark-pro-free-writing-qa-preview