Why this is hard to get right
The Problem With Decision Memos Nobody Reads
Maria is a senior product manager at a 200-person B2B SaaS company. Her team has spent three weeks debating whether to build a native integration or rely on a third-party connector. She has opinions. Her engineering lead has different opinions. The VP of Sales wants it shipped yesterday. The VP of Engineering says six weeks minimum.
Maria's job is to write a decision memo that stops the debate and lands on a path. She opens a blank doc and starts typing — context, options, some bullet points about risks. Two hours later, she has four pages of notes that read more like a journal entry than an executive document. She pastes it into an AI assistant and types: "Turn this into a decision memo."
The result is polished but hollow. It lists the options, but the trade-offs feel made up. The risk section says things like "may impact timelines" without numbers. The recommendation is hedged so heavily that it reads as "it depends." The exec team will ask the same questions in the meeting that the memo was supposed to answer.
The core problem isn't Maria's thinking. It's that she didn't give the AI the structure to reason through her actual situation. A decision memo isn't just a summary of options — it's a structured argument with defined criteria, real constraints, and a defensible recommendation. Without that scaffolding, even good AI produces generic output.
When Maria rebuilt her prompt with a clear role assignment, three explicit options, five evaluation criteria tied to Q2 OKRs, and hard constraints like headcount and a compliance deadline, the result changed completely. The AI produced a one-page memo with a risk table, a ranked comparison of options, and a 30/60/90 execution plan the engineering lead could actually sanction.
The exec team approved it in 15 minutes. Not because the AI was smarter, but because the prompt gave it everything it needed to reason like a senior strategist — not a general-purpose summarizer.
This is the difference between prompting for a document and prompting for a decision. The former produces text. The latter produces clarity.
Common mistakes to avoid
Describing the Situation Instead of Framing the Decision
Many prompts describe background context at length but never name the specific decision being made. The AI then writes a summary, not a memo. Force clarity by stating the decision in one sentence: 'Should we ship Feature X this quarter or delay to Q3?' Everything else in the prompt supports that question — not the other way around.
Omitting the Approval Audience
A memo for your engineering lead reads very differently from one for your CFO. Without naming the audience, the AI defaults to a generic stakeholder, which means the tone, depth, and framing miss the mark. Specify who signs off and what metric they care about most, and the AI calibrates the argument to that reader automatically.
Listing Options Without Defining Evaluation Criteria
Giving the AI three options but no criteria forces it to invent its own comparison framework. You get a comparison that sounds reasonable but doesn't match how your team actually decides. Name your real criteria — revenue impact, churn risk, engineering effort — so the AI evaluates options the same way you would in a room with your exec team.
Leaving Constraints Implicit
Saying 'we have limited resources' is not a constraint. Vague limits produce unrealistic plans — the AI might recommend hiring two engineers you don't have or hitting a deadline that's already passed. State hard numbers: 6 engineers, 10-week runway, SOC2 compliance required. Constraints aren't obstacles; they're the boundaries that make the recommendation credible.
Asking for a Recommendation Without Specifying the Format
Without format instructions, AI-generated memos often run 3-5 pages, bury the recommendation, and skip the risk table. Execs stop reading after page one. Specify exactly what you need: one page, a risk table with likelihood and impact columns, a recommendation paragraph, and a 30/60/90-day plan. Format is not cosmetic — it determines whether the memo gets used.
Skipping the Role Assignment
A prompt that says 'write a decision memo' gives the AI no perspective to reason from. Add a role — 'You are a senior product strategist' — and the AI shifts from neutral summarizer to opinionated analyst. That shift produces bolder recommendations, sharper trade-off language, and memos that actually sound like they came from someone who has made hard calls before.
The transformation
Write a decision memo about whether we should build this feature or not.
You’re a senior product strategist. Write a 1-page decision memo for our exec team about whether we should ship **Feature X** this quarter. 1. **Context:** B2B SaaS, 25-person team, Q2 goal is +15% net revenue retention. 2. **Options (3):** build now, delay 1 quarter, don’t build. 3. **Criteria:** revenue impact, churn risk, engineering effort, time-to-value, support load. 4. **Constraints:** 6 engineers, 10 weeks, must meet SOC2 controls. Use a direct tone. Include a risk table with likelihood/impact, a recommendation, and a 30/60/90-day plan.
Why this works
Role Anchors the Perspective
The After Prompt opens with 'You're a senior product strategist' — a deliberate role assignment that shifts how the AI reasons. Without a role, the AI writes as a neutral summarizer. With one, it adopts the judgment, confidence, and vocabulary of an experienced operator, which produces recommendations that sound defensible rather than hedged.
Defined Options Prevent Scope Drift
The After Prompt names exactly three options — build now, delay 1 quarter, don't build — rather than asking the AI to generate its own list. This keeps the memo focused on the decision at hand. When you let the AI invent options, it often introduces alternatives that aren't feasible or that your team already rejected.
Explicit Criteria Mirror Real Decision-Making
The five criteria listed — revenue impact, churn risk, engineering effort, time-to-value, support load — match the dimensions your exec team actually debates. This forces the AI to evaluate all three options through the same lens, producing a comparison that stakeholders recognize as credible rather than AI-generated boilerplate.
Hard Constraints Produce Feasible Plans
The constraint block — 6 engineers, 10 weeks, SOC2 controls — gives the AI firm boundaries to reason within. Instead of proposing an ideal-world solution, it generates a plan that fits your actual capacity and compliance requirements. This alone prevents the most common executive objection: 'This plan doesn't reflect our real situation.'
Format Specification Drives Usability
The After Prompt specifies a risk table with likelihood/impact, a recommendation, and a 30/60/90-day plan. These aren't stylistic preferences — they're the artifacts decision-makers need to approve a path and assign accountability. Specifying the format means you get a memo you can paste into an email without reformatting.
The framework behind the prompt
The Theory Behind Structured Decision Memos
Decision memos sit at the intersection of two well-established disciplines: structured decision-making and persuasive professional writing. Understanding both explains why prompt structure matters so much for this output type.
Decision-Making Frameworks
The most widely cited framework for structured decisions is the WRAP model (Widening options, Reality-testing assumptions, Attaining distance, Preparing to be wrong), developed by Chip and Dan Heath in Decisive. The model argues that most bad decisions come from narrow framing — considering too few options and evaluating them against unstated criteria. A well-structured decision memo forces the author to widen options and make criteria explicit before a recommendation appears.
Similarly, MCDA (Multi-Criteria Decision Analysis) from operations research provides the theoretical basis for the criteria-and-weighting approach used in professional decision memos. MCDA requires that all options be evaluated against the same criteria, with consistent definitions and scoring — exactly what a structured AI prompt enforces.
Why Memos Fail Without Structure
Research in organizational behavior shows that unstructured decision documents — narrative summaries without explicit options or criteria — produce two consistent failure modes:
- Confirmation bias amplification: Readers interpret vague language to support their existing position, so debate continues after the memo circulates.
- Authority ambiguity: Without a named audience and clear recommendation, it's unclear who should act and on what timeline.
The Role of the Prompt in Mitigating These Failures
The RISEN framework (Role, Instructions, Steps, End goal, Narrowing) is directly applicable here. Assigning a role (senior product strategist), providing instructions (evaluate three options against five criteria), specifying steps (context, comparison, risk table, recommendation, plan), naming the end goal (exec alignment), and narrowing the scope (one page, 10-week constraint) mirrors the RISEN structure almost exactly.
Bloom's Taxonomy is also relevant: generating a recommendation requires the highest cognitive level — evaluation and synthesis — not just recall or comprehension. A prompt that provides criteria and constraints gives the AI the scaffolding to operate at that level rather than defaulting to description.
Prompt variations
You are a principal engineer advising a VP of Engineering. Write a one-page decision memo evaluating whether to build a custom data pipeline or purchase a third-party ETL tool.
Decision: Should we build or buy a data pipeline solution for our analytics platform?
Options:
- Build in-house using our current data engineering team
- Purchase and integrate a third-party ETL vendor
- Hybrid: buy for ingestion, build for transformation
Evaluation criteria: total cost of ownership over 24 months, time to first data in production, vendor lock-in risk, internal maintenance burden, scalability to 10x data volume.
Constraints: 2 senior data engineers available, 8-week deadline before Q3 analytics launch, must support GDPR compliance.
Audience: VP of Engineering and CTO.
Include a side-by-side comparison table, a risk summary for each option, and a clear recommendation with rationale. Keep the memo under 500 words.
You are a marketing strategist. Write a one-page decision memo for a CMO evaluating how to reallocate a $200,000 quarterly marketing budget after a paid search channel underperformed by 35%.
Decision: Where should the $200,000 reallocated budget go for maximum pipeline impact in Q3?
Options:
- Double down on content and SEO to build organic pipeline over 6 months
- Shift budget to field events and ABM targeting top 50 accounts
- Split evenly between LinkedIn paid and content syndication
Criteria: projected pipeline contribution, time to first qualified lead, cost per opportunity, brand authority impact, team capacity to execute.
Constraints: team of 3 marketers, no headcount additions, campaigns must launch within 3 weeks.
Audience: CMO reviewing at the monthly leadership meeting.
Include a risk table, a ranked recommendation with supporting rationale, and suggested 30/60/90-day milestones.
You are a customer success operations analyst. Write a decision memo for the VP of Customer Success evaluating three approaches to handling a new 90-day onboarding policy for enterprise accounts.
Decision: Should we mandate the 90-day onboarding for all enterprise accounts, make it optional, or create a tiered model by contract value?
Options:
- Mandatory for all enterprise accounts above $50,000 ARR
- Optional with proactive outreach from CSMs
- Tiered: mandatory above $100,000, optional below
Criteria: customer satisfaction risk, CSM capacity load, churn reduction potential, time-to-value for new accounts, scalability as we grow from 120 to 200 enterprise accounts.
Constraints: current team of 8 CSMs, no additional hires approved until Q4, new policy must roll out before end of month.
Include a likelihood/impact risk table for each option, a single recommended path, and the top 3 objections the team is likely to raise with responses to each.
You are a chief of staff supporting a CEO. Write a one-page decision memo evaluating three paths to filling a vacant VP of Sales role that has been open for 90 days.
Decision: Should we promote an internal candidate, continue the external search, or bring in a fractional VP while we hire?
Options:
- Promote the current senior sales director into the VP role
- Continue the external search with a new recruiting firm
- Hire a fractional VP of Sales for 6 months while we hire full-time
Criteria: speed to impact, cost, cultural fit risk, team morale effect, revenue continuity during transition.
Constraints: $280,000 base budget for the full-time hire, Q3 pipeline targets require sales leadership in place within 30 days, board review in 6 weeks.
Audience: CEO and board chair.
Include a trade-off table, a recommendation with the strongest counter-argument addressed, and a 30-day action plan for the recommended path.
When to use this prompt
Product Managers Preparing Go/No-Go Decisions
Turn scattered discovery notes into a structured memo that compares options and lands on a recommendation.
Engineering Leaders Evaluating Build vs. Buy
Frame alternatives with effort, timeline, and risk so stakeholders stop debating based on gut feel.
Marketing Teams Prioritizing Campaign Investments
Compare 3 campaign paths using agreed criteria like pipeline impact, lift, and delivery capacity.
Customer Success Leaders Handling Policy Changes
Assess policy options with customer impact, support load, and churn risk before rolling out changes.
Pro tips
- 1
Define the single decision in one sentence so the memo stays focused.
- 2
Specify 3-5 evaluation criteria you already use so the comparison feels credible.
- 3
Add hard constraints like headcount, deadlines, and compliance needs to prevent unrealistic plans.
- 4
Name the approval audience and their top metric so the recommendation matches what they value.
Most decision memos list criteria equally, but real decisions aren't equal. If churn risk matters twice as much as engineering effort, your memo should reflect that.
Add a weighting instruction to your prompt:
'Assign each criterion a weight from 1-3 (3 = most important). Use these weights: revenue impact (3), churn risk (3), engineering effort (2), time-to-value (2), support load (1). Score each option against each criterion and show a weighted total.'
This produces a decision matrix — a quantitative comparison that's much harder for stakeholders to argue against on subjective grounds. It also forces the recommendation to be mathematically traceable.
A few caveats:
- Don't let the matrix override judgment entirely. If Option A scores highest but violates a compliance requirement, note the override and explain it.
- Set weights before you see the scores. Adjusting weights after you see results is rationalization, not analysis.
- Use the matrix as a supporting artifact, not the entire memo. Executives want the recommendation first, the evidence second.
The most effective teams don't write decision memos from scratch each time. They build a standard intake template that feeds directly into the prompt structure.
Here's a lightweight intake form you can adapt:
Decision Intake Form
- Decision in one sentence:
- Options (3 max):
- Evaluation criteria (3-5):
- Hard constraints (budget, headcount, timeline, compliance):
- Approval audience and their top metric:
- Required output format:
- Decision deadline:
When a decision needs a memo, the requester fills out this form. The PM or chief of staff converts it directly into a structured prompt using the decision memo template. The AI generates a first draft in under 2 minutes.
The process removes three common friction points:
- Debate about what options to compare — the form forces alignment before drafting
- Scope creep in the memo — the form limits options to three and criteria to five
- Revision cycles — the first draft already matches exec expectations because the format is preset
Teams that adopt this process typically cut decision cycle time by 30-40% because the debate moves upstream into the intake form, not downstream into the memo review.
A memo written for an internal exec team doesn't automatically work for a board or investor audience. The stakes, vocabulary, and framing shift significantly.
When prompting for board-level decision memos, add these instructions:
- Lead with the financial impact of each option in absolute terms (revenue at risk, EBITDA effect, runway change) — not just percentages.
- Include a 'do nothing' option even if you plan to recommend against it. Boards expect to see the cost of inaction explicitly stated.
- Add a governance note that explains who has authority to approve each option and what vote threshold is required.
- Shift from operational language to strategic language. Instead of 'engineering effort,' say 'capital allocation and time-to-market risk.'
- Reduce the 30/60/90-day plan to a single paragraph. Boards approve direction; management handles execution milestones.
A prompt addition that handles most of this:
'Adapt the language and depth for a board of directors audience. Prioritize financial impact, strategic risk, and governance implications over operational details. Assume the reader understands the industry but not the internal architecture or team structure.'
When not to use this prompt
When This Prompt Pattern Is Not the Right Tool
Don't use a decision memo prompt when the decision is already made. If leadership has a preferred path and you're looking for communication support, use an announcement or change management prompt instead. A memo that appears to evaluate options but steers toward a predetermined answer destroys credibility when stakeholders detect it.
Avoid this pattern for exploratory or ambiguous problems. Decision memos require defined options and clear criteria. If you're still in the discovery phase — trying to understand a problem rather than choose between known solutions — use a research synthesis or problem framing prompt first.
Skip this format for operational or tactical decisions that don't require stakeholder alignment. Deciding which backlog items to pull into a sprint doesn't need a one-page exec memo with a risk table. Match the formality of the output to the stakes of the decision.
This pattern also underperforms for decisions with more than five stakeholders with conflicting mandates. In those situations, the memo becomes a political document, not an analytical one. Consider a facilitated workshop prompt or a stakeholder alignment framework instead.
- Use instead for exploration: research synthesis prompts
- Use instead for communication: announcement or change management prompts
- Use instead for high-conflict situations: stakeholder alignment or RACI prompts
Troubleshooting
The recommendation is too vague or says 'it depends on priorities'
Add a tie-breaking instruction to your prompt. Write: 'If options score similarly across criteria, prioritize the option with the lowest churn risk. Make a single, named recommendation — not a conditional one.' Also check whether your criteria are contradictory. If two options genuinely tie on every dimension, you may need to add one more criterion that differentiates them, such as reversibility or team morale impact.
The risk table uses generic language like 'may impact timelines' instead of specific risks
Seed the risk table with one real example in your prompt. Write: 'Populate the risk table with specific, named risks — for example, SOC2 audit delays, not compliance risk in general. Each risk should name the cause, the affected metric, and the mitigation.' When the AI sees a concrete example format, it mirrors that specificity instead of defaulting to generic placeholders.
The memo runs 4-5 pages and buries the recommendation
Set a hard word limit and move the recommendation to the top. Add to your prompt: 'Keep the total memo under 450 words excluding tables. Open with the recommendation and rationale in the first paragraph. Use headers for Context, Options Comparison, Risk Table, and 30/60/90 Plan. Do not include an executive summary — the opening paragraph serves that purpose.'
The options comparison treats all criteria as equal when some clearly matter more
Assign explicit weights in the prompt. List your criteria with a priority score: 'Weight these criteria: revenue impact (high), churn risk (high), engineering effort (medium), support load (low). Reflect these weights in the comparison — don't treat all factors equally in the final analysis.' This prevents the AI from recommending an option that wins on low-priority factors while losing on the ones that actually matter.
The 30/60/90-day plan includes tasks that aren't feasible given stated constraints
Repeat your constraints in the plan section explicitly. Add: 'When building the 30/60/90-day plan, validate each milestone against these constraints: 6 engineers, 10-week runway, no new hiring approved. Flag any step that requires resources beyond these limits.' Repeating constraints in the format section forces the AI to check feasibility at the planning stage, not after you've already shared the memo.
How to measure success
How to Evaluate the Quality of Your Decision Memo Output
Before sharing the memo with stakeholders, run it through this checklist:
Structure check:
- The recommendation appears in the first paragraph, not the last
- Exactly three options are compared, not more or fewer
- The risk table includes both likelihood and impact columns with specific language
- The 30/60/90-day plan fits within the stated constraints
Content quality signals:
- Specificity: Risks name causes and affected metrics, not generic categories
- Traceability: The recommendation connects explicitly to the highest-weighted criteria
- Feasibility: No milestone in the plan requires resources beyond what you stated
- Audience fit: The tone and vocabulary match the approval audience (exec vs. board vs. operational team)
Red flags that require revision:
- The recommendation is conditional ("it depends on how you weight these factors")
- Risk descriptions use phrases like "may impact" without quantification
- The plan includes steps that violate your stated constraints
- The memo runs more than one page before the risk table
Final test: Could you paste this memo into an email to your executive team right now without editing? If not, identify the specific gap and add it to your prompt for the next iteration.
Now try it on something of your own
Reading about the framework is one thing. Watching it sharpen your own prompt is another — takes 90 seconds, no signup.
Turn your next build-vs-buy or go/no-go decision into a one-page exec memo with a clear recommendation and risk table.
Try one of these
Frequently asked questions
Three options is the standard for executive decision memos. One option is a recommendation disguised as a memo. Two options creates a binary debate. Three gives stakeholders real contrast without decision fatigue. Use: build now, delay, or don't build. Or buy, build, or hybrid. The AI will evaluate each more rigorously when the set is finite and named upfront.
Yes — and it's one of the most valuable uses. Build the prompt honestly with real criteria and constraints, then let the AI validate your reasoning. If the AI reaches the same conclusion, you have a stronger case to present. If it reaches a different one, you either discover a blind spot or you refine your criteria to better reflect your actual priorities. Either way, your recommendation gets sharper.
Narrow to three before prompting. More than four options produces a comparison table that executives won't read. If you have five or more candidates, do a quick pre-screening pass: ask the AI to eliminate options that fail a single hard constraint (budget, timeline, compliance). Use the survivors as your three finalized options in the full decision memo prompt.
Add a dedicated compliance block to the constraints section. Name the specific regulations that apply — SOX, HIPAA, PCI-DSS — and state any review gates the recommendation must pass (legal review, audit committee sign-off). Also instruct the AI to flag any option that introduces regulatory exposure, even if that option scores well on other criteria. This keeps the memo legally defensible.
Hedging usually means the AI lacks enough context to commit. Add this line to your prompt: 'Make a single, unambiguous recommendation. State which option you recommend and why it outperforms the others given the stated criteria and constraints. Do not qualify the recommendation with excessive caveats.' If hedging persists, check whether your criteria are in conflict — competing priorities force the AI to hedge because you haven't told it how to break ties.
Describe the situation with real numbers rather than pasting raw data. Instead of attaching a spreadsheet, say: 'Churn for enterprise accounts is 9% annually. The proposed integration reduces churn by an estimated 2-3 points based on similar implementations.' Synthesized specifics give the AI the signal it needs without requiring it to interpret raw formats, which can introduce errors.
One page is the target for most executive audiences. That's roughly 400-500 words plus a risk table. Specify this explicitly in your prompt — 'Keep the memo under 500 words, excluding the risk table.' If your decision genuinely requires more depth (e.g., a board-level M&A consideration), allow up to two pages but still require the recommendation to appear in the first paragraph.
Yes — the structure is decision-agnostic. Swap the context, options, and criteria blocks for each new decision. Keep the role assignment, the format requirement, and the constraint block as permanent fixtures. Many teams build a decision memo template with these fixed elements and fill in the variable fields from a shared intake form before every major decision review.