{"id":"mkt2-campaign-recap-claim-check","audience":"marketing","category":"fact-checking","track":"marketing","technique":"verification","level":"intermediate","title":"Check six claims in a campaign recap against the data table","free":false,"task":"A colleague wrote a recap of the October webinar campaign, and you must check it before it goes to leadership. Write a prompt for exactly 6 numbered lines, one per claim in order, each starting with \"Confirmed\", \"Wrong\" or \"Not in data\". For Wrong, give the correct figure with its calculation like (x / y = z%) in the same line. After the six lines, add a final line starting \"Summary:\" with the counts of Confirmed, Wrong and Not in data. Under 130 words.","materialLabel":"Data table and recap claims (attached to your prompt automatically)","material":"OCTOBER WEBINAR CAMPAIGN - data\n\nEmails sent: 18,000 | Opens: 5,400 | Clicks: 720 | Registrations: 360 | Attended live: 144 | Spend: $2,700\nRegistration page visits: 1,440 (from email and social)\nLinkedIn posts: 3 | LinkedIn clicks: 410\nPost-event survey: 52 responses.\n\nRECAP CLAIMS (to check)\n1. We sent 18,000 emails.\n2. The open rate was 35%.\n3. About 50% of registrants attended live.\n4. Cost per registration was $7.50.\n5. The webinar generated 40 sales meetings.\n6. LinkedIn drove 410 clicks.\n\nNotes: Open rate = opens / sent. Attendance rate = attended / registered. Cost per registration = spend / registrations. Sales meetings are tracked in the CRM, which isn't included in this table.","rules":[{"type":"numberedItems","count":6},{"type":"maxWords","max":130}],"mistakes":[{"title":"Trusting the recap","why":"The AI nods along with the colleague.","fix":"Say: \"Recompute each figure from the table.\"","checks":["r1"]},{"title":"Guessing the sales meetings","why":"The AI says Confirmed or Wrong without data.","fix":"Say: \"Use 'Not in data' when the table can't show it.\"","checks":["r2"]},{"title":"No correction","why":"A Wrong label with no figure isn't useful.","fix":"Say: \"For Wrong, give the correct figure.\"","checks":["r1"]},{"title":"Merging the claims","why":"The AI groups claims into paragraphs.","fix":"Say: \"One numbered line per claim, in order.\"","checks":["r0","numberedItems"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting and often runs past a word limit.","fix":"Cap each part (for example \"each line under 15 words\") so the total lands well under the limit.","checks":["maxWords"]}],"promptHints":[{"label":"Sets the six numbered lines","patterns":["6 numbered","six numbered","exactly 6","one per claim"]},{"label":"Gives the three labels","patterns":["confirmed","wrong","not in data"]},{"label":"Asks for the correct figure","patterns":["correct figure","give the correct"]},{"label":"Recomputes from the table","patterns":["recompute","from the table","table"]},{"label":"Caps length","patterns":["120 words","under 120"]}],"hint":"Fact-checks grade well with exact labels. Give the three labels, tell the AI to recompute and to give the correct figure for Wrong.","mission":"Check six claims in the October webinar recap against data.","musts":["Confirmed, Wrong, Not in data","Show correct calculation","Summary counts line"],"rubricLabels":["There are exactly 6 numbered lines in the order of the claims, each starting with Confirmed, Wrong or Not in data","Claims 1, 4 and 6 are Confirmed, claim 2 is Wrong with the correct 30% shown as 5,400 / 18,000 = 30%, and claim 3 is Wrong with the correct 40% shown as 144 / 360 = 40%","Claim 5 is Not in data, no line invents a figure outside the table, and a final line starts Summary: with 3 Confirmed, 2 Wrong and 1 Not in data"]}