{"id":"res-cs-contact-reasons-tally","audience":"customer-support","category":"research","technique":"grounding","title":"Report the top contact reasons with counts from the tag export","free":false,"task":"Your support lead wants the top 3 reasons customers contacted you last week for Monday's standup. You want at most 3 bullets in plain text, each a reason with its ticket count and share of the total (like \"Login issues: 120 of 600\"), using the final export after the duplicate cleanup. Use only the export, and don't add causes or fixes.","materialLabel":"The tag export (attached to your prompt automatically)","material":"TICKET TAG EXPORT - Week of Oct 26 (Brightloop meal kits)\n\nPreliminary pull (Oct 31, 9 a.m.): 1,040 tickets\n- Late delivery: 310\n- Missing item: 260\n- Billing question: 190\n- Change or skip a week: 150\n- Recipe question: 70\n- Other: 60\n\nFINAL (Nov 2, after removing 140 duplicate tickets from the email-to-chat bug): 900 tickets\n- Late delivery: 250\n- Missing item: 240\n- Billing question: 180\n- Change or skip a week: 150\n- Recipe question: 40\n- Other: 40\n\nNote from Ops: \"late deliveries are probably the new courier in Ohio\" (not confirmed).","rules":[{"type":"maxBullets","max":3},{"type":"plainText"}],"mistakes":[{"title":"Using the preliminary pull","why":"The first numbers are bigger and come first, so the AI may report them.","fix":"Say: \"Use only the final Nov 2 export.\"","checks":["r0","r1"]},{"title":"No denominator","why":"A count without the total doesn't show how big the problem is.","fix":"Show it: \"Format: Reason: N of 900.\"","checks":["r1"]},{"title":"Adding a cause","why":"The ops note invites the AI to explain late deliveries with an unconfirmed guess.","fix":"Add: \"No causes or fixes.\"","checks":["r2"]},{"title":"Not capping the list","why":"The AI lists everything it finds, so the few points that matter get buried.","fix":"Say: \"at most 3 bullets.\"","checks":["maxBullets"]},{"title":"Getting markdown back","why":"AI defaults to bold labels and headings, which show up as stray symbols in the standup doc.","fix":"Ask for \"plain text, no bold, no headings.\"","checks":["plainText"]}],"promptHints":[{"label":"Uses the final export","patterns":["final","nov(ember)?\\.? 2","900","after (the )?(cleanup|duplicates)"]},{"label":"Asks for counts out of the total","patterns":["of 900","out of","share","count"]},{"label":"Rules out causes","patterns":["no (causes|fixes|guess)","(don't|do not) (explain|guess|add)"]},{"label":"Caps the list","patterns":["\\b(3|three)\\b.{0,15}(bullet|reason)","at most"]},{"label":"Asks for plain text","patterns":["plain","no bold","no markdown","no formatting","no headings"]}],"hint":"Early data pulls often include duplicates. Tell the AI which export is final, ask for counts with the total, and keep guesses out.","rubricLabels":["The 3 reasons are late delivery (250), missing item (240) and billing question (180) from the final export, not the preliminary numbers","Each bullet gives its count out of 900 (the final total), not out of 1,040","Contains only the bullets: no intro, no causes (such as the Ohio courier guess), no suggested fixes"]}