{"id":"pro-pm-research-findings-exec","audience":"product-managers","category":"summarizing","technique":"grounding","title":"Summarize user research findings for an exec with real counts","free":false,"task":"Your head of product wants the top findings from last month's user interviews, in a format she can paste into a board update. You want exactly 3 numbered findings, each one sentence ending with how many participants said it, like (7 of 12), under 60 words, plain text. Use the researcher's corrected count and leave out her personal guesses that no participant said. Write a prompt that gets you exactly that.","materialLabel":"Research summary (attached to your prompt automatically)","material":"User interviews - Fennel meal-kit app - October\nResearcher: Joy Mendez. 12 participants, 45 min each, all current subscribers.\n\nWhat we heard:\n- Skipping a week is hard to find. 9 of 12 said they couldn't find \"skip week\" and some canceled instead. Several said \"I just wanted a break, not to quit.\"\n- Recipes take longer than listed. 7 of 12 said the \"30-minute\" recipes take 45+ minutes.\n- Delivery boxes: 4 of 12 mentioned too much packaging.\n- Price: 3 of 12 said it feels expensive, but most said it's worth it.\n\nMy own hunches (not from participants):\n- I bet younger users would love a TikTok-style recipe video feed.\n- We should probably add a points/rewards program.\n\nCorrection from Joy (Nov 3): \"I miscounted the recipe-time finding. It's 6 of 12, not 7. One participant was talking about a different app.\"\n\nLogistics: participants got a $50 gift card. Recordings are in the research drive. Next round in January.","rules":[{"type":"numberedItems","count":3},{"type":"maxWords","max":60},{"type":"plainText"}],"mistakes":[{"title":"Dropping the numbers","why":"The AI writes 'many users said', which a board can't weigh.","fix":"Add: \"End each finding with how many said it, like (9 of 12).\"","checks":["r0"]},{"title":"Missing the correction","why":"The original count appears first, and the AI may keep it.","fix":"Add: \"Use Joy's corrected count for the recipe-time finding.\"","checks":["r1"]},{"title":"Mixing in hunches","why":"The AI treats the researcher's ideas as findings because they're in the same doc.","fix":"Add: \"Only things participants said, not Joy's hunches.\"","checks":["r2"]},{"title":"Letting the AI pick the layout","why":"The AI writes an exec summary with headings and recommendations.","fix":"Be exact: \"Exactly 3 numbered findings, one sentence each. Plain text.\"","checks":["numberedItems","plainText"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting. Ask for \"under 60 words\" and it often lands well over.","fix":"Cap each part instead: \"each line under 14 words.\"","checks":["maxWords"]}],"promptHints":[{"label":"Says how many findings","patterns":["\\b(3|three)\\b"]},{"label":"Asks for the counts","patterns":["of 12","how many","count","number of (participants|users)"]},{"label":"Uses the correction","patterns":["correct","6 of 12","joy"]},{"label":"Excludes hunches","patterns":["hunch","only (what|things) participants","not .{0,20}opinion","guess"]},{"label":"Sets a word limit","patterns":["\\d+\\s*words","word (limit|count)"]}],"hint":"Execs trust findings with numbers behind them. Ask for the counts, point the AI at the correction, and tell it to skip ideas nobody in the study said.","rubricLabels":["Each of the 3 findings is one sentence ending with a participant count in the form (N of 12)","Covers the skip-week finding (9 of 12) and, if the recipe-time finding is included, uses the corrected count (6 of 12, not 7)","Includes none of the researcher's own hunches (video feed, rewards program) and none of the logistics (gift cards, recordings, next round)"]}