Why this is hard to get right
A Real-World Scenario: When "Summarize Our Quarter" Isn't Enough
Maria is a Director of Operations at a mid-sized SaaS company. Every quarter, she spends the better part of two days pulling data from three dashboards, coordinating with finance and sales, and trying to wrangle it all into a slide deck that leadership will actually read. The last time she handed off a summary, the CEO sent it back with a one-line note: "This is just numbers. What do we do with it?"
She tried using AI to help. She typed: "Summarize our quarterly results for leadership." The output was a generic template with placeholder language — nothing specific to her company's metrics, no prioritized risks, and certainly no recommended actions. She spent another hour editing it from scratch.
The real problem wasn't the AI. It was the prompt.
Maria didn't tell the AI who would read the summary, what data mattered most, what level of technical detail was appropriate, or what kind of decision the summary needed to support. Without that context, the AI produced the safest possible output — which was also the least useful one.
When Maria restructured her approach, she specified the audience (non-technical leadership), the exact metrics that mattered (revenue, churn, acquisition cost, product usage), the format (one page, no walls of text), and the desired outcome (three recommended actions). She also set the role — a senior business analyst — so the AI's tone reflected sound judgment, not a data dump.
The difference was immediate. The AI produced a focused, confident summary that called out three critical insights and flagged an emerging churn risk she had almost buried in footnotes. Her leadership team read it before the meeting. The CEO asked two questions instead of sending it back for revision.
This is the professional reality of quarterly executive summaries: the stakes are high, the audience is impatient, and vague AI outputs create more work, not less. A well-crafted prompt isn't a luxury — it's the difference between a summary that drives decisions and one that collects dust.
The good news is that the structure is learnable and repeatable. Once you know which elements to include — role, audience, metrics, format, and outcome — you can produce executive-ready summaries in a fraction of the time, every single quarter.
Common mistakes to avoid
Omitting the Audience's Technical Level
When you don't specify whether leadership is technical or non-technical, the AI defaults to a middle ground that satisfies no one. A CFO and a CTO need very different language for the same revenue data. Always state the audience's background explicitly so the AI calibrates depth, vocabulary, and the amount of context it provides around each metric.
Listing No Specific Metrics
Saying 'include our key metrics' forces the AI to guess. It often produces generic placeholders like 'revenue' and 'growth' without connecting them to your actual numbers or thresholds.
- Name the exact metrics (e.g., CAC, NRR, churn rate, ARR)
- Provide current values where possible so the AI can frame insights, not just categories
Skipping the Output Format
Without a format constraint, AI tends to produce long, meandering prose. Executive summaries need to be scannable in under two minutes. Specify a page limit, use of headers, bullet points, or a numbered structure. A vague prompt produces a wall of text that leadership won't read before the meeting.
Forgetting to Request Recommended Actions
Most users ask for a summary of what happened but forget to ask for guidance on what to do next. Leadership doesn't just want a report — they want to make a decision. Add an explicit instruction to conclude with 2-4 prioritized recommendations to make the summary immediately actionable.
Not Specifying the Reporting Period and Data Sources
Ambiguous timeframes lead to ambiguous analysis. Without knowing you mean Q4 FY2024 versus the trailing 90 days, the AI makes assumptions. State the exact quarter, fiscal year, and which data sources (CRM, product analytics, finance systems) so the AI structures the analysis around the right scope.
Assigning No Analytical Role to the AI
Prompts that skip role assignment get neutral, cautious summaries that hedge every claim. Telling the AI to act as a senior business analyst signals that you want interpretation and judgment — not just a restatement of the data. This single addition consistently raises the quality and confidence of the output.
The transformation
Summarize our quarterly results for leadership.
**Act as a senior business analyst.** Produce a **one-page executive summary** of our **Q4 performance** for a non-technical leadership team. 1. Highlight key metrics: revenue, growth rate, acquisition cost, churn, and top product usage trends. 2. Identify 3 critical insights and why they matter. 3. Flag any risks or emerging patterns. 4. Use a confident, concise tone. 5. End with 3 recommended actions for the next quarter.
Why this works
Role Assignment Raises Output Quality
The phrase "Act as a senior business analyst" at the start of the After Prompt sets the AI's analytical posture. It tells the model to apply judgment, not just organize data. This produces insights with a confident, professional tone rather than a neutral list of facts — which is exactly what leadership expects from a quarterly summary.
Audience Specificity Controls Depth
The After Prompt explicitly targets "a non-technical leadership team." This single phrase prevents the AI from defaulting to jargon-heavy analysis. It calibrates vocabulary, determines how much context to provide around each metric, and ensures the summary is readable without a background in data science or engineering.
Named Metrics Eliminate Generic Output
Listing "revenue, growth rate, acquisition cost, churn, and top product usage trends" directly in the prompt tells the AI exactly what to analyze. Without this, the AI produces placeholder categories. With it, the output organizes itself around your actual business drivers and avoids irrelevant metrics that dilute the summary.
Numbered Structure Forces Completeness
The five-step numbered list in the After Prompt — from key metrics through to recommended actions — acts as a structural contract. The AI follows each instruction in sequence, which prevents omissions like missing risk flags or a summary that ends without clear next steps. Structure in the prompt becomes structure in the output.
Actionable Ending Drives Decision-Making
Closing the After Prompt with "End with 3 recommended actions for the next quarter" transforms a recap into a decision-support tool. This instruction ensures the summary doesn't stop at describing what happened — it tells leadership what to do about it, which is the primary job of any executive communication.
The framework behind the prompt
The Strategy Behind Executive Summaries
Executive summaries exist at the intersection of information design and decision science. Their purpose isn't to report everything that happened — it's to reduce cognitive load for decision-makers who are managing multiple priorities and limited reading time. Research in organizational communication consistently shows that senior leaders make faster, more confident decisions when information is presented in pyramid structure: conclusion first, evidence second, detail available on request.
This principle is formalized in the Minto Pyramid Principle, developed by Barbara Minto at McKinsey, which argues that effective executive communication answers the key question first and supports it with grouped, logically ordered arguments. Most AI-generated summaries violate this by defaulting to chronological or categorical structure — the way data is collected rather than the way decisions are made.
A second relevant framework is SCQA (Situation, Complication, Question, Answer), also from Minto's work. Strong quarterly summaries follow this arc: here's where we were (Situation), here's what changed or challenged us (Complication), here's the strategic question that raises (Question), and here's what we recommend (Answer). When your prompt includes metrics, risk flags, and recommended actions in sequence, you're effectively encoding the SCQA structure into the AI's output.
The audience calibration element of a strong prompt draws on principles from technical communication theory, which distinguishes between expert, practitioner, and lay audiences. Non-technical leadership requires contextual translation — converting data points into business meaning — which the AI only performs reliably when you explicitly specify the audience's background.
Finally, the role-prompting technique (assigning the AI a professional persona like "senior business analyst") draws on research showing that persona framing shifts language models toward more domain-appropriate vocabulary, reasoning depth, and output confidence. This is consistent with findings published in prompt engineering research from Stanford and DeepMind teams studying how context priming affects large language model output quality.
Understanding these frameworks helps you make deliberate prompt choices rather than guessing what to include.
Prompt variations
Act as a senior sales analyst preparing a quarterly business review for a VP of Sales and CFO.
Produce a one-page QBR summary for Q3 covering:
- Pipeline performance: total pipeline value, average deal size, and win rate versus Q2.
- Revenue attainment: actual versus quota by region and segment.
- Top 3 deals won and top 3 deals lost — with one-sentence explanations for each.
- Churn or contraction events that affected net revenue retention.
- Three priorities for Q4 with specific targets attached to each.
Tone: Direct and data-driven. Assume the audience will interrogate every number. Use plain language. No filler sentences.
Act as a senior product analyst.
Summarize Q2 product performance for a cross-functional leadership team that includes non-technical stakeholders.
- Report on feature adoption rates for the three features released this quarter.
- Summarize NPS movement and the top three recurring themes from customer feedback.
- Flag any features with adoption below 15% and explain the likely cause.
- Connect product performance to revenue impact where data supports it.
- Recommend two roadmap adjustments based on this quarter's evidence.
Format: Use clear headers for each section. Keep the full summary under 400 words. Write for a reader who has 5 minutes, not 30.
Act as a senior marketing analyst.
Produce a Q1 marketing performance summary for a CMO and CEO who care about pipeline contribution and brand momentum, not campaign mechanics.
- Summarize total leads generated, MQL-to-SQL conversion rate, and cost per MQL versus target.
- Identify the top two channels by ROI and the one channel that underperformed expectations.
- Highlight any brand or awareness metrics that moved meaningfully (share of voice, organic traffic, branded search volume).
- Note any external factors — seasonality, competitive moves, market shifts — that affected results.
- Recommend one budget reallocation and two campaign priorities for Q2.
Tone: Confident and concise. Lead with the insight, follow with the evidence. Avoid marketing jargon.
Act as an operations analyst presenting to a COO.
Summarize Q4 operational performance with a focus on efficiency, cost control, and process reliability.
- Report on cost per unit or cost per transaction versus Q3 and versus annual target.
- Summarize SLA adherence rates and any incidents that breached thresholds.
- Highlight one process improvement that delivered measurable results this quarter.
- Flag two operational risks heading into Q1 — with probability and potential impact noted.
- Close with three specific operational priorities for Q1, each with an owner and success metric.
Format: One page. Use bullet points under each numbered section. Write for a COO who reads fast and questions everything.
When to use this prompt
Marketing Directors
Create quarterly summaries that highlight campaign performance and give leadership clear insights to guide budget decisions.
Product Managers
Summarize product adoption, customer feedback, and release impact to support roadmap planning.
Sales Leaders
Translate pipeline, win rates, and segment performance into leadership-ready insights for quarterly business reviews.
Operations Teams
Prepare efficiency, cost, and process performance summaries for executive planning sessions.
Pro tips
- 1
Add your exact metrics so the AI focuses on what matters most.
- 2
Define your leadership audience to control tone and complexity.
- 3
Include timeframe and data sources to avoid misinterpretation.
- 4
Specify actions or recommendations you want included to support decision-making.
The After Prompt on this page focuses on a single quarter in isolation. For senior leadership audiences, comparative framing consistently produces stronger analysis because it answers the most common executive question: compared to what?
To add comparative context, extend your prompt with one of these additions:
- "Compare each metric against Q3 performance and against the same quarter last year. Note which direction each metric moved and by what percentage."
- "Where targets were set at the start of the quarter, state the target alongside the actual result and calculate the variance."
- "Benchmark our churn rate and CAC against published industry medians for B2B SaaS companies at our ARR tier."
The third option requires the AI to draw on its training data, so treat those benchmarks as directional rather than authoritative. For internal comparisons (quarter-over-quarter or year-over-year), provide the historical figures directly in the prompt so the AI doesn't interpolate.
A practical tip: paste a small data table directly into the prompt. AI models handle structured input well, and a 5-row comparison table often produces sharper analysis than a paragraph describing the same numbers.
The core structure of this prompt travels well across industries, but the metric sets and risk categories differ significantly. Here are sector-specific adjustments:
SaaS and Subscription Businesses Prioritize NRR, GRR, logo churn, and expansion revenue. Add: "Separate new ARR from expansion ARR in the revenue section."
E-commerce and Retail Focus on GMV, average order value, return rate, and inventory turnover. Add: "Flag any seasonal demand patterns that explain metric movement."
Professional Services Emphasize utilization rate, revenue per billable hour, project margin, and pipeline coverage ratio. Add: "Note any client concentration risks if a single client represents more than 20% of quarterly revenue."
Healthcare Organizations Shift the metric set to patient volume, average revenue per encounter, payer mix, and days in accounts receivable. Add: "Flag any compliance-related operational changes that affected performance."
In each case, the structural bones of the prompt remain the same: role, audience, specific metrics, format, and recommended actions. Only the metric vocabulary and risk categories change.
Running this prompt without preparation often produces a second editing pass. Use this pre-flight checklist to avoid that:
Data readiness
- Do you have final (not provisional) figures for all named metrics?
- Do you have the prior-quarter baseline for comparison?
- Have you confirmed which data source is the system of record for each metric?
Audience clarity
- Can you name the specific individuals or roles who will read this summary?
- Do you know their tolerance for detail and their primary decision concern this quarter?
Constraints and format
- Have you set a word or page limit in the prompt?
- Do you know whether the summary will be read on screen, printed, or presented verbally?
Outcome focus
- Are you clear on the one decision you want leadership to make after reading this?
- Have you told the AI what constraints apply to the recommended actions (budget, headcount, timeline)?
If you can answer yes to all of these before running the prompt, your first output will require minimal revision.
When not to use this prompt
When This Prompt Pattern Is Not the Right Tool
This prompt works well for structured, data-driven quarterly reviews. But there are specific situations where it won't serve you well:
When the data isn't finalized. Running this prompt on provisional or estimated figures produces a polished-looking summary built on shaky numbers. Leadership trust erodes fast when a confident AI output turns out to contain unverified data. Wait for closed numbers.
When the summary requires confidential judgment calls. If your quarterly summary involves sensitive personnel decisions, M&A activity, or board-level strategic pivots, AI output isn't appropriate for a first draft. These situations require human authorship and review before any AI involvement.
When the audience expects a narrative, not a structured report. Some cultures and leadership styles prefer flowing narrative memos over bullet-pointed summaries. In those contexts, the numbered structure in this prompt will produce output that feels mismatched with expectations. Adapt the format instruction accordingly.
- For narrative-first cultures, try: "Write this as a flowing memo, not a bulleted report."
- For very short C-suite briefings, reduce scope to a single page of 3 bullets maximum.
When you need legally reviewed language. Any summary that will be shared externally, filed with regulators, or included in investor communications needs legal review. Don't use AI output as the final draft in those contexts.
Troubleshooting
The AI summary reads like a data dump — no clear insights or narrative thread
Add an explicit instruction to lead each section with the insight, not the number. For example: 'For each metric, state what the result means for the business before listing the data point. The insight comes first, the evidence follows.' This forces the AI to interpret before it reports, which is what distinguishes an executive summary from a spreadsheet export.
Recommended actions are too vague (e.g., 'invest more in top channels')
Tell the AI to attach a measurable outcome and a timeframe to each recommendation. Add: 'Each recommendation must include a specific success metric and a target completion date within Q1.' If actions are still vague, follow up with: 'Make recommendation 2 more specific — what exact action should the team take in the first two weeks of Q1?'
The output is written in technical language that non-technical leadership won't follow
Reinforce the audience constraint with a plain-language rule. Add: 'Assume the reader has no background in data analysis or technical operations. Every metric must include a one-sentence plain-language explanation of what it measures and why it matters.' You can also add: 'Avoid all acronyms unless you define them on first use.'
The risk section is either missing or overly alarming
Separate risk identification from risk framing in your prompt. Add: 'Identify 2 emerging risks. For each, rate severity as low, medium, or high and state one early indicator to monitor.' This forces structured, calibrated risk language rather than the binary choice the AI makes between ignoring risks entirely or overstating them.
The summary is too long and the AI ignores word count instructions
Replace relative length instructions ('one page', 'concise') with an absolute word ceiling. Write: 'This summary must not exceed 400 words total. If needed, cut supporting detail before cutting insights or recommendations.' After the first output, follow up: 'This is [X] words. Reduce to 400 words by trimming the metric descriptions in section 1 and section 2.'
How to measure success
How to Evaluate the Quality of Your AI-Generated Summary
Before you share any AI-generated executive summary, run it through this quality checklist:
Clarity and audience fit
- Could a non-technical leader read this in under 3 minutes and understand the core findings?
- Does it avoid undefined acronyms or unexplained technical references?
Completeness
- Are all five elements present: key metrics, critical insights, risks, tone, and recommended actions?
- Does each section contain a specific finding — not a placeholder or generic statement?
Actionability
- Do the recommended actions name a specific next step, not a vague direction?
- Can a reader immediately assign ownership to each recommendation?
Structural discipline
- Does the summary stay within your specified length constraint?
- Does each section lead with the insight before the supporting evidence?
Risk calibration
- Are risks framed with evidence, not speculation?
- Is each risk flag accompanied by a suggested early-warning indicator?
If the output fails two or more of these checks, revise the prompt rather than editing the output manually. Prompt revision produces better results faster than downstream editing.
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 Q4 data into a leadership-ready executive summary — in one structured prompt.
Try one of these
Frequently asked questions
Start with the metrics you do have and tell the AI to note gaps rather than invent numbers. Add a line like: "Where data is unavailable, flag it as a gap rather than estimating." This prevents fabricated figures and gives you a useful draft you can fill in as your data comes in. You can run the prompt again with complete data before distribution.
Yes, but you'll get better results by adding a segmentation instruction. After your metric list, add: "Organize findings by business unit: [Unit A], [Unit B], [Unit C]." Also specify whether you want a consolidated view first or unit-by-unit breakdowns. Without this, the AI tends to average across units, which hides meaningful variance between high- and low-performing segments.
Add an explicit word ceiling. Replace 'one-page summary' with 'a summary of no more than 350 words.' If it's still too long, follow up with: 'Reduce this to 300 words by cutting supporting detail and keeping only the primary insight for each section.' Word limits are more reliable constraints than page-length references, which the AI interprets loosely.
The AI produces vague recommendations when it lacks context about your constraints. Add one sentence like: "Recommendations should account for a Q1 budget freeze and a team headcount of 12." Constraints force specificity. You can also tell the AI to attach a named metric or deadline to each recommendation to make them immediately evaluable.
The core structure works for board reporting, but you'll need two adjustments. First, shift the role to 'senior executive communicator' rather than business analyst. Second, add: 'Prioritize strategic narrative over operational detail. Flag only risks with material financial or reputational impact.' Board audiences want strategic signal, not operational noise.
Add a calibration instruction after your risk request: "Frame risks as emerging patterns with supporting evidence. Avoid speculative or worst-case language unless the data clearly supports it." This produces measured risk flags that leadership can act on rather than alarming statements that create panic without a clear response pathway.
Yes. Change the framing to 'a mid-quarter performance snapshot' and reduce the metric list to 2-3 leading indicators. Replace the recommended actions section with: 'Identify one early corrective action if current trends continue through end of quarter.' This keeps the output proportional to the amount of data available at the halfway point.