Resource · AI Prompts · Product Leadership
AI Prompts for
Product Leaders
Three production-tested prompts for high-stakes PM work — from deep research decisions to executive storytelling to business case synthesis. Copy, adapt, use.
Prompt 01
Deep Research: Buy vs. Build Decision
Use when: evaluating whether to build a web component in-house or adopt an open-source / vendor solution
The Prompt
# Context I am a Senior Product Manager evaluating whether to [build in-house / adopt open-source / license a vendor solution] for [specific component or capability]. Our constraints: - Team: [number of engineers, their expertise] - Timeline: [desired delivery window] - Scale: [number of products / users affected] - Tech stack: [frameworks, design system context] - Non-negotiables: [accessibility, compliance, brand consistency, etc.] # Task Conduct a structured buy vs. build analysis. For each option (build, open-source, vendor), evaluate: 1. **Total Cost of Ownership** over 3 years — include build, maintain, upgrade, and opportunity cost 2. **Time to value** — realistic delivery timeline including integration and adoption 3. **Strategic fit** — does this build platform leverage or create one-off technical debt? 4. **Risk profile** — what are the top 3 failure modes for each option? 5. **Hidden costs** — what do teams typically underestimate in each scenario? # Output format - Structured comparison table (option × dimension) - A recommended option with a clear rationale - 3 questions I should pressure-test with my engineering lead before deciding - One question that would change your recommendation if the answer were different
Why this prompt works
Most buy vs. build conversations stall at surface-level cost comparisons. The hidden costs and pressure-test questions force the AI to surface the uncomfortable truths your team isn't saying out loud — vendor lock-in, migration debt, the engineer who becomes the sole maintainer. The final question ("what would change your recommendation") is a forcing function for intellectual honesty.
Strategic Decision Platform Product Works with: Claude, GPT-4, Gemini
Prompt 02
Executive Presentation Builder
Use when: presenting a strategy, investment ask, or initiative update to VP / C-suite audience
The Prompt
# Context I need to build an executive presentation for [audience: e.g. VP of Engineering + CFO] on [topic: e.g. scaling our design system with AI tooling]. Their priorities are: [e.g. cost efficiency, engineering velocity, platform leverage] Their concerns are likely: [e.g. ROI timeline, team capacity, risk of AI reliability] The decision I need: [e.g. approval of $500K AI tooling investment] Time available: [e.g. 15 minutes + 10 minutes Q&A] Key data I have: [paste your metrics, outcomes, business context] # Task Design a presentation structure that leads executives to the decision, not through the work. 1. **Opening frame** — what is the single most important thing they need to understand in the first 60 seconds? 2. **Slide structure** — recommend a 5–7 slide arc with a title and one-line purpose per slide 3. **The ask** — how should I frame the investment request to minimize friction? 4. **Anticipated objections** — list the top 3 objections and draft a one-sentence response to each 5. **Leave-behind** — what is the one data point they will remember after the meeting? # Constraints - No jargon. Every term should be explainable in one sentence. - Each slide must answer: "So what does this mean for us?" - The narrative arc should feel like a story, not a status update.
Why this prompt works
Executives are pattern-matching for risk and opportunity — they're not reading your slides. The "lead to the decision, not through the work" instruction breaks the habit of building presentations that tell the story of how hard you worked, rather than what the audience needs to decide. The leave-behind question forces ruthless prioritization.
Executive Communication Storytelling Investment Ask Works with: Claude, GPT-4
Prompt 03
Business Case Data Synthesizer
Use when: you have raw data from multiple sources and need to find the signal that makes a product case compelling
The Prompt
# Context I am building a business case for [initiative]. I have data from the following sources: [Source 1: e.g. CSAT survey — paste key findings or data] [Source 2: e.g. usage analytics — paste key findings or data] [Source 3: e.g. stakeholder interviews — paste themes] [Source 4: e.g. support tickets — paste top categories] My audience cares most about: [e.g. cost savings, developer velocity, revenue risk] # Task Synthesize this data to find the strongest business case. Specifically: 1. **The headline** — one sentence that captures the core argument. Should be specific, quantified if possible, and falsifiable. 2. **Cross-source patterns** — which themes appear in 2+ data sources? These are your highest-confidence claims. 3. **The cost of inaction** — what happens if we don't act? Frame this in terms my audience cares about. 4. **Weak points** — where is the data thin or inconclusive? Flag these so I can either address or de-emphasize them. 5. **Recommended narrative arc** — a 3-act structure: here's the problem, here's what we know, here's what I'm proposing. 6. **The one metric** — if I could only track one number to prove this investment worked, what should it be and why? # Important Do not invent data. If something can't be supported by what I've shared, say so. Flag assumptions clearly.
Why this prompt works
Business cases fail when they cherry-pick data rather than synthesize it. The cross-source pattern instruction forces triangulation — claims that appear in only one source are hypotheses, not evidence. The "cost of inaction" frame is often the most persuasive element a PM forgets to include. And the explicit instruction not to invent data keeps the output honest.
Pro tip: Run this prompt once to get the structure, then run it again with the actual document draft and ask: "Where is my narrative weakest? What would a skeptical CFO push back on first?"
Business Case Data Synthesis Stakeholder Communication Works with: Claude, GPT-4, Gemini