Turn raw data into an executive briefing with this reusable AI prompt
Why this prompt matters
Raw tables rarely persuade anyone on their own. A structured briefing helps teams spot the real signal faster, avoid talking past each other, and make decisions before a meeting turns into manual spreadsheet narration.
What we use it for
You have a weekly leadership review in 30 minutes and only a rough CSV export from your product, marketing, or operations dashboard.
Prompt
Role: Act as a senior analytics strategist and chief of staff who turns messy operational data into clear executive briefings. Context: I will give you [RAW DATA OR CSV], [METRIC DEFINITIONS], [BUSINESS CONTEXT], [TIME WINDOW], and [TARGET AUDIENCE]. The data may be incomplete, inconsistently formatted, or missing useful labels. Your job is to identify the signal, not just restate the table. Task: Analyze the input and produce a decision-ready executive briefing that explains what changed, what matters, what is uncertain, and what action should be considered next. Constraints: - Use only the information provided. If something important is missing, say so explicitly. - Distinguish observed facts from hypotheses or interpretations. - Quantify changes wherever possible (absolute and percentage change when relevant). - Highlight only the most material trends, anomalies, risks, and opportunities. - Do not flood the reader with every metric. - If the data quality is weak, include a short "data confidence" note. - If useful, ask up to [NUMBER OF FOLLOW-UP QUESTIONS] clarifying questions first, but still provide a provisional briefing. - Keep the tone concise, executive, and specific. Output Format: 1. Executive Summary (5-7 bullet points) 2. Key Trends Table - Metric - What changed - Why it likely matters 3. Anomalies and Red Flags 4. Likely Drivers and Caveats 5. Recommended Actions for [TARGET AUDIENCE] 6. Data Confidence Note Input: - Raw data / CSV: [PASTE DATA HERE] - Metric definitions: [DEFINE THE METRICS] - Business context: [DESCRIBE THE PRODUCT / TEAM / BUSINESS] - Time window: [E.G. LAST 7 DAYS VS PRIOR 7 DAYS] - Target audience: [E.G. CEO, COO, VP GROWTH] - Decision needed: [WHAT DECISION THIS SHOULD SUPPORT] - Follow-up question limit: [E.G. 3]
Result
<p><strong>Executive Summary</strong></p><p>- New trial signups fell 14 percent week over week, but the larger issue is lower activation after signup, not just top-of-funnel volume.</p><p>- Activation dropped from 42 percent to 34 percent, which means the business lost conversion efficiency at the point where value should become visible to a new user.</p><p>- Paid acquisition spend increased 11 percent while cost per activated user rose 32 percent, suggesting recent campaign mix is sending weaker traffic or landing-page intent is mismatched.</p><p>- Expansion revenue grew 9 percent and churn stayed mostly flat, so this does not look like a broad product failure. The pressure is concentrated in new-user onboarding.</p><p>- Support tickets mentioning import errors increased 38 percent in the same period. That is the strongest operational clue in the dataset.</p><p>- Immediate focus should shift from buying more traffic to verifying whether onboarding friction is suppressing activation.</p><p><strong>Key Trends Table</strong></p><p><strong>Trial signups:</strong> 8,420 → 7,241 (-14.0%). This matters because top-of-funnel volume is softening, but the decline is not severe enough by itself to explain the full revenue risk.</p><p><strong>Activation rate:</strong> 42% → 34% (-8 points / -19.0%). This is the most material movement in the dataset because it directly affects pipeline quality for paid conversion.</p><p><strong>Paid media spend:</strong> $118,000 → $131,000 (+11.0%). Spend rose while activation quality fell, reducing efficiency.</p><p><strong>Cost per activated user:</strong> $33.40 → $44.10 (+32.0%). This indicates deteriorating unit economics in acquisition.</p><p><strong>Support tickets tagged “import error”:</strong> 211 → 291 (+37.9%). This likely points to onboarding friction that deserves immediate product and support review.</p><p><strong>Anomalies and Red Flags</strong></p><p>The support-ticket jump is too closely timed with the activation decline to ignore. The most plausible explanation is that a product, integration, or UX issue is interrupting account setup. A secondary possibility is traffic quality degradation from a new paid campaign, but current data supports the onboarding-friction hypothesis more strongly.</p><p><strong>Likely Drivers and Caveats</strong></p><p>The dataset does not include landing-page breakdowns, device segmentation, or release notes, so the root cause is not confirmed. However, the numbers are directionally consistent with a setup bottleneck rather than pure demand weakness.</p><p><strong>Recommended Actions</strong></p><p>1. Audit onboarding flows and import logs from the last 7 days.<br/>2. Pull campaign-level activation quality before increasing paid budget.<br/>3. Ask support to classify the import-error spike by integration, browser, and account type.<br/>4. Prepare a follow-up readout within 24 hours with segmented activation data.</p><p><strong>Data Confidence Note</strong></p><p>Confidence is moderate. The trend is clear, but root-cause certainty is limited because the export lacks segmentation and product release context.</p>
<p>Most teams do not struggle to collect data. They struggle to turn a raw export into something a leadership team can act on quickly. That is the gap this prompt is built to close. Instead of asking an AI model to simply “analyze this CSV,” it gives the model a clear role, business context, decision goal, and output structure so the result reads like an executive briefing rather than a stream of observations.</p> <p>The core thesis is simple: analytics prompts get better when you stop treating the model like a calculator and start treating it like an analyst writing for a specific audience. Executives do not need every number repeated back to them. They need the signal: what changed, what matters, what is uncertain, and what decision is now on the table.</p> <h2>What this prompt is designed to do</h2> <p>This prompt works best when you have a rough table, spreadsheet export, dashboard copy-paste, or KPI snapshot and need to brief a manager, founder, client, or operations lead. It pushes the model to organize information into five useful layers: a short executive summary, the most important trends, the anomalies worth investigating, the likely business implications, and a practical action list.</p> <p>That structure matters because most raw-data prompts fail in one of two ways. They either produce a wall of descriptive statistics with no decision value, or they jump too quickly into recommendations without being clear about which numbers are facts and which statements are interpretation. This prompt forces the model to separate evidence from inference. That makes the output easier to trust and easier to challenge.</p> <h2>Why the prompt uses Role, Context, Task, Constraints, and Output Format</h2> <p>The <strong>Role</strong> section tells the model to behave like a senior analytics strategist and chief of staff, not a generic chatbot. That changes tone and prioritization. You want the model to think about tradeoffs, audience sensitivity, and decision usefulness.</p> <p>The <strong>Context</strong> section gives it the operating conditions: the company, the business model, the audience, the reporting window, and the definitions behind the numbers. Without that, even a capable model can misread a conversion dip, overreact to seasonality, or miss the difference between a one-off campaign and a structural change.</p> <p>The <strong>Task</strong> section defines the actual job. It is not “summarize the data.” It is “produce a decision-ready executive briefing.” That wording matters. It tells the model to optimize for clarity and actionability rather than exhaustiveness.</p> <p>The <strong>Constraints</strong> section is where the quality jump happens. It explicitly tells the model to quantify changes, call out missing information, distinguish observed facts from hypotheses, and avoid inventing certainty. In real reporting workflows, that is the difference between something you can forward and something you need to rewrite from scratch.</p> <p>Finally, the <strong>Output Format</strong> gives the reader a reusable template. A good prompt does not just create one strong answer; it creates a consistent reporting pattern your team can use every week.</p> <h2>Where this prompt is especially useful</h2> <p>This is a strong fit for weekly business reviews, growth-team recaps, customer support trend summaries, marketplace ops reporting, finance snapshots, or product KPI check-ins. It is also useful when the data is incomplete. The prompt instructs the model to flag blind spots and still produce a provisional read, which is often exactly what teams need before a meeting.</p> <p>It is less useful when you need formal statistical testing, financial controls, or domain-specific regulatory interpretation. In those cases, the prompt is still valuable as a first-pass briefing layer, but it should not replace an analyst, accountant, or operator who owns the numbers.</p> <h2>How to get better results from it</h2> <p>Three edits improve the output immediately. First, define the audience precisely: “COO,” “VP Growth,” or “client services director” is better than “leadership.” Second, include metric definitions when names are ambiguous. Third, tell the model what decision is pending. If the reader is choosing between budget reallocation, a campaign pause, or a product rollback, the recommendations become much sharper.</p> <p>The best part of this prompt is that it scales down as well as up. A founder can paste one rough table and get a clear memo. A larger team can wire the same structure into recurring reporting workflows. In both cases, the value is the same: less time turning exports into prose, and better odds that the right person notices the right signal before the next meeting starts.</p>