{"id":"mgr-weekly-metrics-post","audience":"managers","category":"data","technique":"examples","title":"Post this week's support numbers in last week's format","free":false,"task":"Every Monday you post your support team's numbers in the team channel, in the same 4-line format as last week's post (shown in the material). Write this week's post in exactly that format, plain text, under 30 words, nothing else. Apply the ops notes, since they change two of the numbers, and use the median first-reply time like last week.","materialLabel":"Last week's post and this week's data (attached to your prompt automatically)","material":"LAST WEEK'S POST (our team format)\nSupport wk of Oct 27\nClosed: 198 (up 6)\nFirst reply: 3.4h median\nShout-out: Marcus, 14 customer thank-yous\n\nTHIS WEEK - export from HelpHub, week of Nov 3\nTickets closed: 212\nFirst reply time: median 2.9h, average 4.1h\nReopened: 9\nCustomer thank-yous by agent: Jen 16, Priya 11, Marcus 9, Tariq 6\n\nOPS NOTES (Mon)\n- 12 of this week's closed tickets were spam that the filter auto-closed. Don't count them.\n- 7 of Jen's thank-yous came from the internal IT team testing the survey. They don't count.\n- HelpHub upgrade is scheduled for Nov 18, expect 30 minutes of downtime.\n\nRANDOM\n- Tariq is back from leave, welcome back!\n- Holiday coverage sign-up sheet is in the shared drive.","rules":[{"type":"maxWords","max":30},{"type":"plainText"}],"mistakes":[{"title":"Not pointing at the example","why":"The AI invents its own report layout, with headings, sections and emojis.","fix":"Add: \"Match last week's post exactly: same 4 lines, same order, same labels.\"","checks":["r0","plainText"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting. Ask for a total word limit and it often lands over it.","fix":"Set the shape instead: \"Only the 4 lines, nothing else.\"","checks":["maxWords"]},{"title":"Ignoring the ops notes","why":"The AI copies the raw export, so spam tickets and test thank-yous inflate the numbers.","fix":"Add: \"Apply the ops notes before you work out any number.\"","checks":["r1","r2"]},{"title":"Picking the wrong number","why":"The export lists both a median and an average, and the AI may grab the average.","fix":"Add: \"Use the median first-reply time, like last week.\"","checks":["r1"]}],"promptHints":[{"label":"Points at last week's post","patterns":["last week","same format","match","example|template"]},{"label":"Applies the ops notes","patterns":["ops notes","spam","internal|test","notes"]},{"label":"Uses the median","patterns":["median"]},{"label":"Sets a word limit","patterns":["\\d+\\s*words","word (limit|count)"]},{"label":"Asks for plain text","patterns":["plain","no bold","no markdown","no formatting","no headings"]}],"hint":"You already have the perfect example: last week's post. Tell the AI to copy its format exactly, and to fix the raw numbers with the ops notes first.","rubricLabels":["Matches last week's format: the same 4 lines in the same order (Support wk of Nov 3 / Closed: N (up or down N) / First reply: Xh median / Shout-out: Name, N customer thank-yous) and nothing else","Closed is 200 (212 minus the 12 spam tickets), shown as up 2 from last week's 198, and first reply is 2.9h median (not the 4.1h average)","The shout-out goes to Priya with 11 customer thank-yous (Jen drops to 9 once the 7 internal test thank-yous are removed)"]}