{"id":"res-off-newsletter-hr-stat-labels","audience":"office","category":"fact-checking","technique":"verification","title":"Label each newsletter stat against the HR quarterly report","free":false,"task":"You're proofing the company newsletter's \"By the Numbers\" box before it goes out. You want exactly 6 numbered lines, one per stat, each labeled Matches, Wrong, or Not in report, and each Wrong line must give the report's figure. Use only the Q3 HR report as the source, in plain text, with nothing else.","materialLabel":"The HR report and the newsletter draft (attached to your prompt automatically)","material":"Q3 2026 HR REPORT - Ferris & Lowe Engineering (all offices)\nHeadcount on Sept 30: 284\nNew hires in Q3: 22\nVoluntary turnover, Q3: 3.1%\nEmployees who completed safety training: 91%\nAverage tenure: 6.4 years\nOpen requisitions on Sept 30: 17\nNote: Q2 headcount was 268; Q2 new hires 15.\n\n-----\nNEWSLETTER DRAFT - \"By the Numbers\" box\n1. We're now 284 strong!\n2. We welcomed 15 new teammates this quarter.\n3. Turnover held at just 3.1%.\n4. Almost everyone (98%) finished safety training.\n5. Our people stay an average of 4.6 years.\n6. Employee satisfaction is up 12 points.","rules":[{"type":"numberedItems","count":6},{"type":"plainText"}],"mistakes":[{"title":"No fixed labels","why":"The AI writes a paragraph of comments, so the editor can't scan which stats to fix.","fix":"Give them: \"Label each Matches, Wrong, or Not in report.\"","checks":["r2"]},{"title":"Missing the sneaky ones","why":"Q2's 15 hires sits in the report, so a quick read passes them.","fix":"Add: \"Check every number against the Q3 figures only, not Q2.\"","checks":["r1"]},{"title":"Calling missing stats wrong","why":"The AI treats the satisfaction stat as false, when the report just doesn't cover it.","fix":"Say: \"If the report doesn't include it, label it Not in report.\"","checks":["r0"]},{"title":"Not saying how many","why":"The AI decides the length itself and usually returns more lines, one per stat than you can use.","fix":"Be exact: \"exactly 6 numbered lines, one per stat.\"","checks":["numberedItems"]},{"title":"Getting markdown back","why":"AI defaults to bold labels and headings, which show up as stray symbols in the newsletter tool.","fix":"Ask for \"plain text, no bold, no headings.\"","checks":["plainText"]}],"promptHints":[{"label":"Gives exact labels","patterns":["matches","not in (the )?report","label"]},{"label":"Names the source","patterns":["only.{0,30}report","report.{0,20}(only|source|truth)","q3"]},{"label":"Asks for the right figure","patterns":["report('s)? figure","correct (number|figure)","give the"]},{"label":"Says exactly how many","patterns":["\\b(6|six)\\b.{0,25}(line|stat)","exactly \\d+"]},{"label":"Asks for plain text","patterns":["plain","no bold","no markdown","no formatting","no headings"]}],"hint":"Stats can be slightly off in ways that sound right. Give the AI exact labels, make the Q3 report the only judge, and ask for the report's number when one is wrong.","rubricLabels":["Labels stats 1 and 3 Matches, and stat 6 (satisfaction) Not in report","Labels stats 2, 4 and 5 Wrong with the report's figures (22 new hires, not Q2's 15; 91%, not 98%; 6.4 years, not 4.6)","Outputs only the 6 labeled lines using the labels Matches, Wrong or Not in report: no intro, no rewritten box, no suggestions"]}