{"id":"acct-sample-test-exceptions-no-projection-lines","audience":"accounting","category":"finance-work","track":"finance","technique":"constraints","level":"advanced","title":"Report sample exceptions without projecting to the population","free":false,"task":"You're reporting internal audit sample results to the controller. Write a prompt for plain text with lines starting \"Sample tested:\", \"Exceptions:\", \"Exception rate:\", \"Exception dollars:\", \"Share of sample value:\" and \"Not concluded:\". Count an exception even if it was corrected later. The Not concluded line says that projecting to the full population is for the controller to decide, and gives no projected number. Show the math. Under 120 words.","materialLabel":"Source material (attached to your prompt automatically)","material":"INVOICE SAMPLE TEST - October 2026\n\nPopulation: 400 sales invoices, $2,000,000 total\nSample tested: 25 invoices, $250,000 total (selected by the controller)\n\nExceptions found:\n- E1: INV-3318, $1,200, no signed purchase order on file\n- E2: INV-3351, $8,000, billed in October for services delivered Nov 2 (wrong period)\n- E3: INV-3377, $450, price differs from the contract rate; a credit memo was issued on Nov 4 and the customer was made whole\n\nTesting rule: an item is an exception if it fails at the test date, even if it was later corrected. Exception dollars are the invoice amounts of the failed items.\n\nNote from a staff member: \"3 of 25 is 12%, so about 48 of the 400 invoices must have problems.\" The controller has not reviewed this note and has not approved any projection method. The auditors will receive the sample results on Nov 18.","rules":[{"type":"plainText"},{"type":"maxWords","max":120}],"promptHints":[{"label":"Names the six lines","patterns":["sample tested","exceptions","exception rate","exception dollars","not concluded"]},{"label":"Counts corrected items","patterns":["even if","corrected","later"]},{"label":"Bans projecting","patterns":["do not project","no projection","controller"]},{"label":"Asks for the math","patterns":["show the math","math","/"]},{"label":"Caps length","patterns":["\\d+\\s*words","under \\d+"]}],"mistakes":[{"title":"Dropping the corrected item","why":"The AI excludes E3 because it was fixed.","fix":"Say: \"Count exceptions that failed at the test date, even if corrected.\"","checks":["r1"]},{"title":"Projecting to the population","why":"The AI says about 48 invoices are wrong.","fix":"Add: \"Do not project; Not concluded says that is the controller's decision.\"","checks":["r2"]},{"title":"Missing the percentage math","why":"The AI gives rates without equations.","fix":"Add: \"Show 3 / 25 and $9,650 / $250,000.\"","checks":["r0","r1"]},{"title":"Markdown extras","why":"The AI adds bold labels, headings and a closing note.","fix":"Say: \"Plain text, no bold. Output only the lines asked for.\"","checks":["plainText"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting and runs past word limits.","fix":"Ask for fewer words than the real cap and keep each part short.","checks":["maxWords"]}],"hint":"A sample result describes the sample. Ask for the counts and dollars, and keep the projection out.","rubricLabels":["The six lines start Sample tested:, Exceptions:, Exception rate:, Exception dollars:, Share of sample value: and Not concluded: in that order; Sample tested shows 25 invoices and $250,000, Exceptions shows 3 and Exception rate shows 3 / 25 = 12%","Exception dollars shows $1,200 + $8,000 + $450 = $9,650, counting E3 even though it was corrected, and Share of sample value shows $9,650 / $250,000 = 3.9% (3.86%)","Not concluded says projecting to the 400-invoice population is the controller's decision, and no line gives a projected count or dollar amount such as 48 invoices"]}