Maya Patel·
My one-page decision memo prompt. Takes a folder of messy research notes and returns options, evidence quality, and a recommendation the exec team actually reads to the end.
Converts raw research notes into a one-page decision memo with graded evidence, honest uncertainty, and a recommendation separated from the analysis.
Evidence-to-Decision Memo
You are a research analyst writing a one-page decision memo for a time-poor executive audience. Your job is to compress without distorting — the memo must survive a hostile reader who checks it against the sources.
Decision question: {{decision_question}}
Audience and what they already know: {{audience}}
Raw notes, excerpts, and data (unordered):
{{research_notes}}
Rules:
1. **Answer first.** Open with the recommendation in two sentences maximum, including the confidence level (high / moderate / low) and the single factor most likely to change it.
2. **Options table.** Every realistic option including 'do nothing', each with: one-line description, strongest point for, strongest point against, rough cost or effort. No option gets a strawman treatment — write each 'for' case as its advocate would.
3. **Grade the evidence.** For each claim that carries weight in the analysis, tag it: [strong] multiple independent sources or direct data, [moderate] single credible source, [weak] anecdote, vendor claim, or inference. If my notes contradict each other, surface the contradiction explicitly instead of silently picking a side.
4. **Separate analysis from recommendation.** The analysis section must be written so that a reader could reasonably arrive at a different conclusion; the recommendation section then argues for one.
5. **Uncertainty paragraph.** What we do not know, why it matters, and which unknown is cheapest to resolve — with a concrete next step and rough effort estimate.
6. **Length discipline.** 450 words maximum, no filler phrases ('it is important to note'), no bullet points inside the recommendation itself.
If the notes are insufficient to support any recommendation at moderate confidence, say exactly that and specify the minimum additional evidence needed.
Ergebnisse
## Memo: Should we consolidate on a single survey platform?
**Recommendation (moderate confidence):** Consolidate on Platform A over 2 quarters, keeping Platform B licenses only for the CX team's transactional surveys until the API migration path is verified. Most likely to change this: if A's per-response pricing at our projected volume exceeds the enterprise quote by more than 20%, the cost case inverts.
**Options**
| Option | For (strongest) | Against (strongest) | Effort |
|---|---|---|---|
| Consolidate on A | 70% of teams already there; deepest analysis features [strong — usage export] | Weak transactional/API surveys vs B [moderate] | ~2 quarters |
| Consolidate on B | Best API + automation; CX team's workflows depend on it [strong] | Would migrate 5× more users than option A; analysts rate its reporting last [moderate — internal survey, n=14] | 3+ quarters |
| Keep both, formalize split | Zero migration cost; each tool where it's strongest | Duplicate spend continues (~$61k/yr [strong — invoices]); data stays siloed, which triggered this review | Low |
| Do nothing | None found beyond deferral | Shadow licenses growing 15%/yr [moderate — single procurement pull] | — |
**Analysis.** Usage data favors A on breadth: 210 active builders vs 38 on B [strong]. But raw counts overweight casual use — B's 38 users run the automated NPS pipeline that touches every customer [strong]. The contradiction in the notes: the procurement summary claims B's contract auto-renews in November, while the vendor email states March. This changes the decision timeline materially and is currently unresolved. Feature-gap claims for A's API come from the vendor roadmap [weak] — treat as unverified. A hostile reader will note the internal satisfaction survey (n=14, analysts only) is directional at best; I have graded it accordingly.
**Recommendation case.** Migration cost scales with user count for casual users but with workflow complexity for pipelines. A absorbs casual users cheaply now; B's pipeline stays put until A's API claims are tested against a real transactional survey — a two-week pilot, not a roadmap slide.
**Uncertainty.** We do not know actual renewal dates (cheapest to resolve: one email to both vendors, this week), A's true API capability [pilot, ~2 weeks, one engineer], or volume pricing at consolidation scale [quote request]. The renewal date is the binding constraint — if November is right, the pilot must start this month.
*Word count: 398.*
Modell: o3
24 Likes15 SavesScore: 18
2 Kommentare
Anna Hofmann·
Separating analysis from recommendation is the move. Most memos argue from sentence one and the reader rightly stops trusting them by paragraph two.
Julia Moser·
Used this for a tooling decision memo today. The evidence grading forced me to admit half my 'facts' were one vendor webinar. Uncomfortable and correct.