Table of Contents
Here's a painful truth: most people use AI at 5% of its capacity. They type vague one-liners and get vague one-liner answers back. Then they conclude AI isn't useful.
The difference between a forgettable AI output and one that saves you 40 hours of work comes down to one thing: the prompt. Not the model. Not the subscription tier. The prompt.
I've tested thousands of prompts across ChatGPT, Claude, Gemini, and open-source models. These 15 are the ones I keep coming back to because they consistently produce outputs I'd actually pay for. I'm sharing the full reasoning behind each one so you can adapt them, not just copy them.
Why 99% of AI Prompts Fail
Before the prompts themselves, you need to understand why most prompts produce garbage. There are three failure modes:
- Vagueness -- "Write me a marketing plan" gives the AI zero constraints. It defaults to the most generic, middle-of-the-road answer possible.
- No context -- The AI has no idea about your business, audience, or goals. It's guessing. And it guesses badly.
- No format specification -- You get a wall of text when you needed a table. Or bullet points when you needed a narrative. The AI doesn't know what shape your answer should take.
Every prompt in this guide addresses all three. That's what makes them work.
The Prompt Framework That Changes Everything
I use a framework I call RCOF: Role, Context, Objective, Format. Every great prompt hits all four. Here's what each does:
- Role -- Tell the AI WHO it is. "You are a conversion copywriter with 15 years of experience" produces wildly different output than "write me copy."
- Context -- Feed it YOUR specific situation. Industry, audience, constraints, what you've already tried.
- Objective -- What exactly should the output accomplish? Not what it should be. What it should DO.
- Format -- Table, numbered list, email draft, code block. Be specific about structure.
Now let's apply it.
Business & Strategy Prompts
1. The Competitive Moat Analyzer
You are a strategy consultant who has advised 50+ startups through Series A to exit.
I run a [YOUR BUSINESS TYPE] serving [YOUR AUDIENCE]. My main competitors are [COMPETITOR 1], [COMPETITOR 2], and [COMPETITOR 3].
Analyze my competitive landscape and identify:
1. Three defensible moats I can build in the next 90 days
2. The single biggest vulnerability each competitor has
3. An "unfair advantage" play that none of them are doing
4. A specific pricing strategy that positions me as the premium option
Be contrarian. I don't want generic advice like "build a community." I want specific, executable moves with estimated effort and impact.
Why it works: The "be contrarian" instruction and the "not generic advice like..." constraint forces the model to bypass its default safe answers. Adding the consultant framing with a specific track record activates deeper strategic reasoning. The 90-day constraint makes outputs actionable, not theoretical.
2. The Revenue Stream Architect
You are a business model designer specializing in digital businesses.
I currently make money from [YOUR CURRENT REVENUE SOURCE]. My audience is [SIZE] people who are [DESCRIPTION]. My average transaction value is [AMOUNT].
Design 5 additional revenue streams I could launch within 60 days, each one requiring less than $500 to start. For each one include:
- Revenue model (one-time, recurring, usage-based)
- Realistic month-1 revenue projection with assumptions stated
- The single biggest risk and how to mitigate it
- A 2-week launch plan with daily actions
Rank them by effort-to-revenue ratio, best first.
Why it works: The constraints ($500, 60 days, daily actions) prevent pie-in-the-sky suggestions. Asking for assumptions behind projections forces intellectual honesty. The ranking instruction means the best idea is always first.
3. The Customer Interview Script Generator
You are a product researcher trained in the Jobs To Be Done framework and the Mom Test methodology.
I'm building [PRODUCT/SERVICE] for [TARGET AUDIENCE]. I want to validate whether people will pay for this before I build it.
Create a 15-minute customer discovery interview script that:
- Never asks "would you buy this?" or any hypothetical purchase question
- Focuses entirely on past behavior, current pain, and money already spent
- Includes 3 "drill-down" questions for each topic that dig into specific examples
- Has a natural conversational flow (not an interrogation)
- Ends with a commitment ask that doesn't feel salesy
Add annotations explaining the psychology behind each question.
Why it works: Referencing specific methodologies (JTBD, Mom Test) grounds the output in proven frameworks instead of generic interview advice. The "never ask hypothetical" constraint is the key insight that separates useful customer research from useless data.
Writing & Content Prompts
4. The Hook Factory
You are a viral content strategist who has written hooks for posts with 10M+ impressions.
Topic: [YOUR TOPIC]
Audience: [WHO'S READING]
Platform: [Twitter/LinkedIn/Newsletter/Blog]
Write 20 hooks for this topic using these proven frameworks:
1. Contrarian take (challenge a common belief)
2. Specific number + unexpected outcome
3. "I studied X and here's what I found"
4. Before/after transformation
5. Question that creates an open loop
For each hook, write a brief note on WHY it would stop the scroll. Kill any hook that sounds like every other post on the platform. I want zero generic motivational language.
Why it works: The framework categories force variety. The "kill any hook that sounds generic" instruction acts as a quality filter. The "why it stops the scroll" annotation means you can learn the principles, not just copy the hooks.
5. The Authority Article Outliner
You are an SEO content strategist and long-form writer who has ranked 200+ articles on page 1 of Google.
Keyword: [YOUR TARGET KEYWORD]
Search intent: [INFORMATIONAL/COMMERCIAL/TRANSACTIONAL]
My unique angle: [WHAT MAKES YOUR PERSPECTIVE DIFFERENT]
Create a detailed article outline that:
- Answers the searcher's actual question in the first 100 words
- Has H2/H3 structure optimized for featured snippets
- Includes 5 original data points or examples I should research/create
- Has a "this is what everyone else gets wrong" section
- Naturally integrates [YOUR PRODUCT/SERVICE] as a solution without being pushy
- Ends with a section that makes readers want to bookmark and share
Mark which sections should include images, tables, or comparison charts.
Why it works: The "unique angle" input prevents the AI from producing the same outline as every other top-ranking article. The featured snippet optimization and the visual content markers make this outline directly actionable for publishing, not just brainstorming.
6. The Email Sequence Writer
You are a direct-response email copywriter who has generated $10M+ in email revenue.
Product: [YOUR PRODUCT]
Price: [PRICE]
Audience pain: [MAIN PAIN POINT]
Transformation: [DESIRED OUTCOME]
Write a 5-email welcome sequence. Each email must:
- Have 3 subject line options (one curiosity-based, one benefit-based, one urgency-based)
- Be under 200 words
- Tell a mini-story or share a specific proof point
- End with ONE clear CTA (not multiple)
- Build on the previous email's narrative
The sequence arc should be: Connect → Educate → Prove → Overcome objections → Convert.
Write in a tone that's casual but not cringy. Like texting a smart friend, not a corporate newsletter.
Why it works: The 200-word constraint prevents bloat. The sequence arc ensures strategic flow, not random emails. The tone instruction ("texting a smart friend") is specific enough for the AI to calibrate appropriately. Three subject line options per email means you can A/B test immediately.
Coding & Technical Prompts
7. The Code Reviewer
You are a senior software engineer with 15+ years of experience doing code reviews at companies like Stripe and Vercel.
Review this code for:
1. Bugs or edge cases that will break in production
2. Security vulnerabilities (SQL injection, XSS, auth bypass, etc.)
3. Performance issues that won't matter at 100 users but will crash at 10,000
4. Readability improvements (naming, structure, documentation)
For each issue:
- Severity: Critical / Warning / Suggestion
- Show the problematic line
- Show the fixed version
- Explain the "why" in one sentence
Don't comment on things that are fine. Only flag actual issues.
```
[YOUR CODE HERE]
```
Why it works: The severity classification prevents noise. The "don't comment on things that are fine" instruction eliminates filler praise. Asking for both the problem and the fix means you get actionable output. The scale-aware performance check catches issues that testing misses.
8. The Architecture Decision Document
You are a systems architect designing for a startup that needs to ship fast but not accumulate crippling tech debt.
I need to build [FEATURE/SYSTEM]. The constraints are:
- Team size: [NUMBER] developers
- Timeline: [WEEKS/MONTHS]
- Expected load: [USERS/REQUESTS]
- Must integrate with: [EXISTING SYSTEMS]
Create an architecture decision document covering:
1. Three viable approaches (don't just pick one -- show me the tradeoffs)
2. A comparison table: complexity, time-to-ship, scalability ceiling, maintenance burden
3. Your recommended approach with specific reasoning
4. The migration path if we outgrow this approach
5. The three decisions we'll regret most and how to make them reversible
Be opinionated. I want your actual recommendation, not a neutral overview.
Why it works: Asking for three approaches prevents premature commitment. The comparison table forces structured thinking. "Decisions we'll regret" is an unusually effective framing that surfaces risks other approaches miss. "Be opinionated" overrides the AI's tendency to hedge.
Marketing & Sales Prompts
9. The Landing Page Conversion Optimizer
You are a conversion rate optimization specialist with experience optimizing pages that do $1M+/year.
Here is my landing page copy:
[PASTE YOUR COPY OR URL DESCRIPTION]
Audit this page for:
1. Headline: Does it communicate the primary benefit in under 8 words?
2. Social proof: Is it specific enough? (numbers > vague testimonials)
3. CTA: Is there one clear action? Is the button text benefit-oriented?
4. Objection handling: What are the top 3 reasons someone would NOT buy, and does the page address them?
5. Urgency/scarcity: Is it genuine or fake? (fake hurts conversion long-term)
For each issue, give me:
- The current version
- Your improved version
- The expected conversion impact (small / medium / significant)
Then give me the single highest-impact change I should make today.
Why it works: The specific audit criteria mean nothing gets missed. The "expected conversion impact" ranking helps you prioritize. The "single highest-impact change" at the end gives you immediate actionability even if you don't implement everything.
10. The Objection Destroyer
You are a sales psychologist who understands why people don't buy even when they need and want the product.
My product: [PRODUCT] at [PRICE]
Target buyer: [DESCRIPTION]
Most common objection: [WHAT THEY SAY]
For this objection, create:
1. The REAL reason behind the stated objection (they say price, they mean trust)
2. Three response frameworks: Logic, Story, Question
3. A specific email or message I can send when I hear this objection
4. A way to preemptively address this on my sales page so I never hear it
5. The "nuclear option" -- a guarantee or offer structure that makes the objection irrelevant
Don't give me manipulative tactics. Give me responses that are honest and actually help the buyer make the right decision for THEM.
Why it works: The "real reason vs. stated reason" framing gets past surface-level objection handling. Three response frameworks give you versatility for different situations. The "no manipulation" constraint produces responses that build long-term trust instead of one-time sales.
Analysis & Research Prompts
11. The Market Research Synthesizer
You are a market research analyst who has done due diligence for venture capital firms.
I'm entering the [MARKET/NICHE] market. I need a rapid-fire market brief covering:
1. Market size: TAM, SAM, SOM (with your reasoning, not just numbers)
2. Growth trajectory: Is this market expanding, mature, or declining? Why?
3. The top 5 players and their actual positioning (not just what they say -- what they DO)
4. The underserved segment everyone is ignoring
5. Three market signals that indicate where this market is heading in 18 months
6. The "entry wedge" -- the specific sub-market I should own first before expanding
Be specific. If you don't have exact data, give me directional estimates and label them clearly. I'd rather have calibrated estimates than vague hand-waving.
Why it works: The VC due diligence framing activates rigorous analytical reasoning. Asking the AI to label uncertain data as estimates prevents hallucinated statistics. The "entry wedge" concept is a specific strategic framework that produces actionable insight, not just information.
12. The Decision Matrix Builder
You are a decision scientist who helps executives make high-stakes decisions under uncertainty.
I need to decide between:
- Option A: [DESCRIBE]
- Option B: [DESCRIBE]
- Option C: [DESCRIBE] (if applicable)
Context: [YOUR SITUATION, CONSTRAINTS, PRIORITIES]
Build me a decision matrix that:
1. Identifies the 7 most important criteria (weighted by importance)
2. Scores each option honestly (1-10 with justification)
3. Highlights which option wins overall AND which wins in different scenarios
4. Identifies the "regret minimization" choice (which will I regret least in 5 years?)
5. Names the one piece of information that would change the recommendation
Don't be neutral. Tell me what you'd actually choose and why.
Why it works: Weighted criteria prevent equal-weighting bias. Scenario analysis shows you how different priorities change the answer. The "regret minimization" frame (borrowed from Bezos) adds a time dimension most analysis misses. Asking for the "one piece of information" identifies your biggest knowledge gap.
Advanced Power-User Prompts
13. The System Prompt Builder
You are an AI prompt engineer who designs system prompts for production AI applications.
I want to build an AI assistant that:
- Serves this role: [DESCRIPTION]
- For this audience: [WHO USES IT]
- Primary use case: [WHAT THEY'LL ASK IT TO DO]
- Tone: [FORMAL/CASUAL/TECHNICAL/etc.]
- Constraints: [WHAT IT SHOULD NEVER DO]
Create a production-ready system prompt that:
1. Defines the assistant's identity and expertise clearly
2. Sets behavioral boundaries without being overly restrictive
3. Includes 3 example interactions (input/output pairs)
4. Has a fallback instruction for out-of-scope questions
5. Is under 500 tokens (efficiency matters in production)
Test the prompt against these edge cases:
- User asks something outside the scope
- User tries to jailbreak or override instructions
- User asks a valid but ambiguous question
Why it works: This is a meta-prompt: using AI to build better AI instructions. The 500-token constraint prevents bloated system prompts that waste compute. Edge case testing built into the prompt means you don't have to test separately.
14. The Week-in-One-Hour Planner
You are a productivity strategist who has helped executives 10x their output without working more hours.
Here's everything on my plate this week:
[LIST ALL YOUR TASKS, PROJECTS, MEETINGS, DEADLINES]
My available deep work hours this week: [NUMBER]
My energy peaks at: [MORNING/AFTERNOON/EVENING]
My biggest priority this quarter: [ONE THING]
Create a week plan that:
1. Identifies the 3 tasks that move the needle most (everything else is maintenance)
2. Time-blocks my deep work on needle-moving tasks during peak energy
3. Batches all low-energy tasks into a single "admin block"
4. Identifies 2 things I should delegate or delete entirely
5. Builds in a 30-minute Friday review to assess what actually shipped
Be ruthless. If something doesn't serve the quarterly priority, flag it as a distraction even if it feels urgent.
Why it works: The quarterly priority filter prevents busyness masquerading as productivity. Energy-aware scheduling is something most planning tools ignore. The "delegate or delete" instruction forces hard choices you'd otherwise avoid. The Friday review creates a feedback loop.
15. The "Teach Me Like I'm Building It" Prompt
You are an expert in [TOPIC] who teaches by building, not by lecturing.
I want to deeply understand [SPECIFIC CONCEPT]. My current level: [BEGINNER/INTERMEDIATE/ADVANCED]. I learn best by doing, not reading theory.
Teach me this concept by:
1. Give me a 2-sentence intuition (what is this REALLY about, in plain English)
2. Walk me through building a simple but complete example from scratch
3. At each step, explain the "why" before the "how"
4. After the example works, show me how to break it intentionally (understanding failure = understanding deeply)
5. End with 3 "what if" exercises that push me slightly beyond what you just taught
Do NOT use analogies like "think of it like a factory" unless they're genuinely precise. Bad analogies create worse misunderstanding than no analogy at all.
Why it works: "Building, not lecturing" creates hands-on learning. The "break it intentionally" step is the secret weapon -- you understand something far deeper when you know how it fails. The bad analogy prohibition prevents the AI's worst habit: using imprecise metaphors that feel helpful but mislead.
Want 500+ Prompts Like These?
The AI Prompt Bible has 500+ field-tested prompts for business, writing, coding, marketing, and more. Every single one uses the RCOF framework and has been tested across models.
$27
Get the AI Prompt Bible
Or get 10 prompts free first
How to Adapt These Prompts
Copying prompts is a starting point, not the finish line. Here's how to make any prompt work better for your specific situation:
- Always fill in the brackets with real data. The more specific your inputs, the more specific the outputs. "B2B SaaS" is better than "business." "$4,500 ACV with 14-day trial" is better than "B2B SaaS."
- Iterate on the first output. The first response is rarely the final answer. Say "This is good. Now make it more specific to [your niche]" or "The third point is weak. Give me 5 alternatives."
- Stack prompts. Use the output of one prompt as input for the next. Run the Competitive Moat Analyzer first, then feed those moats into the Landing Page Optimizer.
- Add your voice. Paste a sample of your existing writing and say "Match this tone and voice." AI adapts remarkably well to stylistic examples.
The goal isn't to use AI as a replacement for thinking. It's to use it as a force multiplier for YOUR thinking. These prompts give the AI enough structure to be brilliant while leaving enough space for your unique context to shape the output.
That's the difference between a prompt that wastes your time and one that saves you 40 hours.
Free: 10 AI Prompts That Print Money
The top 10 prompts from our Prompt Bible. No email required.
Download Free