Verified against ChatGPT · 2026-08-12
Compress a stack of research notes into an executive briefing an exec will actually read to the end
Turns raw research notes into a one-page executive briefing that leads with the decision-relevant takeaway and pushes supporting detail into an appendix, instead of a chronological recap of everything you found.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Compress the research notes below into an executive briefing for someone who has five minutes and needs to leave with a clear takeaway, not a chronological account of how the research unfolded. RAW RESEARCH NOTES Competitor A raised prices 15% last quarter with no visible churn spike. Competitor B tried the same and lost 8% of accounts within two months. Our support ticket volume correlates more with feature gaps than price in exit surveys. WHO IS READING THIS The VP of Product, who is under pressure from the board to show a credible path to margin improvement this quarter. DECISION AT STAKE Whether to raise prices 10% across the board next quarter. MY OWN RECOMMENDATION (if I have one) I'm currently leaning toward raising prices but only for new customers, not existing ones. Open with the bottom line, not the background — the first two sentences should state what this means for Whether to raise prices 10% across the board next quarter., before any explanation of methodology or process. Do not lead with "we conducted research into X"; lead with what was found and what it implies. Reorganize the notes by decision relevance, not by the order I researched them in — group findings under what supports the decision, what argues against it, and what's still uncertain, since a chronological recap is the format research notes naturally come in but never the format a time-pressed reader needs. Cut anything from the notes that's interesting but doesn't bear on Whether to raise prices 10% across the board next quarter. — a briefing is not obligated to use everything I gathered. If I gave you my own recommendation, state clearly whether the compressed evidence actually supports it, partially supports it, or points a different direction — do not silently reshape the evidence to flatter a recommendation I've already committed to; an executive briefing that quietly confirms whatever the requester already believed is worse than no briefing at all. Push any supporting detail an exec might ask about in the room — the numbers behind a claim, a caveat, a source note — into a labeled appendix rather than the main body, so the main body stays skimmable in under a minute while the detail is still one scroll away if someone asks a follow-up. OUTPUT FORMAT 1. Bottom line (2-3 sentences). 2. Supports / against / uncertain, as short bullets. 3. Recommendation verdict, if one was given. 4. Appendix: detail bullets, clearly separated from the main body.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The instruction to open with the bottom line rather than background targets a specific and very common failure of model-generated summaries: GPT-5.1's default compression of long notes tends to preserve the original narrative order — methodology, then findings, then implications — because that's the shape most source material comes in, but that order is precisely backwards for an executive reader who needs the implication first and only wants the methodology if they ask. Reorganizing by decision relevance instead of chronology forces genuinely different findings that happen to be about the same topic into an argument structure (supports, against, uncertain) rather than a list, which is what actually helps someone make a call rather than just informs them. The instruction to check the given recommendation against the compressed evidence rather than assume it's correct directly counters a subtle form of sycophancy: when a draft already states what the requester wants to conclude, a model asked to summarize supporting research tends to shape the compression toward confirming that conclusion, quietly downgrading contradicting notes to minor caveats — naming this risk explicitly and requiring an honest verdict (supports, partially supports, or contradicts) breaks that default and makes the briefing actually useful for catching a wrong call before it's made in front of a board. Pushing supporting detail to a clearly labeled appendix rather than cutting it entirely solves the real tension in executive communication between being skimmable and being defensible when someone asks a follow-up question in the room — the appendix keeps both properties without forcing a choice between them.
What you get back
Bottom line: the price increase carries real churn risk based on Competitor B's outcome, and our own exit-survey data suggests the actual driver of churn is feature gaps, not price — raising prices without addressing that gap first repeats the pattern that hurt Competitor B rather than Competitor A. Supports: no visible churn from Competitor A's increase. Against: Competitor B's near-identical move lost 8% of accounts. Uncertain: whether our product's feature-gap profile is closer to A's or B's at the time of their price changes. Recommendation verdict: partially supported — new-customer-only pricing avoids the existing-base churn risk seen in Competitor B's case, but the underlying feature-gap issue should be addressed independently of the pricing decision.
Verified against
ChatGPT GPT-5.1 · 2026-08-12
Changelog
- 2026-08-12 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
