Over the past year, I’ve been running dozens of experiments with different AI agents and custom workflows. The magic often stays in the instructions to the LLM, which can make or break your results, whether summarizing market research or drafting a go-to-market plan.
Crafting an effective prompt rarely happens on the first try. It is often a frustrating, throw-your-laptop-out-the-window kind of process, with multiple passes—tweaking wording, reordering instructions, refining examples—before landing on something that delivers.
The real challenge? I often don’t know the reasoning or constraints buried in a prompt I didn’t write myself. Without visibility into its internal logic, I end up chasing blind alleys.
That’s why I began reverse-engineering complex prompts, especially those used in business and management contexts, such as those found in the Disciplined Entrepreneurship framework (including market segmentation, ideal customer profiles, GTM strategies, and more). In the next section, I’m sharing my full protocol so you can see exactly how it works and apply it to your own AI workflows.
Breaking Down The Process
Introduction (Role, Goal, Input, Workflow)
This is my standard approach to most complex prompts. It establishes clear context and expectations upfront to prevent scope creep and ensure the AI understands its specific function as a prompt engineer rather than a general assistant. The explicit goal statement prevents the model from wandering into tangential analysis, while the input specification creates a standardized interface that users can reliably follow.
## ROLE
You are an expert prompt engineer specializing in reverse-prompt engineering, iterative optimization, and evaluation.
## GOAL
You will analyze output examples, reconstruct the original prompt that generated them, and create an improved version through systematic analysis and testing, using the workflow below.
## INPUT
You will receive three output examples (<OUTPUT_1>, <OUTPUT_2>, <OUTPUT_3>) generated by an unknown prompt and an input (<MINIMUM_VIABLE_INPUT>) that is expected to generate those outputs.
## WORKFLOW
Follow the seven-phase protocol below. Think step by step. Keep detailed reasoning internal, expose only concise rationales and final deliverables. Never reveal full chain-of-thought processes.
Phase 1 (Deep Output Analysis)
We force a systematic examination of surface and hidden patterns before jumping to conclusions. Without this structured analysis, reconstruction attempts often miss subtle constraints or formatting requirements that distinguish good prompts from mediocre ones. The tabular deliverable format ensures consistent, comparable analysis across all examples.
### Phase 1: Deep Output Analysis
Analyze each example for:
- Content structure (headers, lists, paragraphs, formatting)
- Tone, style, and linguistic characteristics
- Format specifications and constraints
- Subject matter depth and technical level
- Implicit instructions evident in outputs
- Common themes across all examples
**Deliverable:** Analysis table with columns: *Feature*, *Evidence (≤20 words)*, *Applies to Example(s)*
Phase 2 (Prompt Reconstruction)
Creates a baseline hypothesis that can be tested and improved upon. This prevents the model from starting with a blank slate and provides a reference point for measuring improvement. The structured elements ensure no critical prompt components are overlooked during reconstruction.
### Phase 2: Prompt Reconstruction
Reconstruct the probable original prompt including:
- Role declarations and persona
- Context setup and background
- Input specifications
- Formatting directives
- Style requirements
- Constraints and limitations
**Deliverable:** <ORIGINAL_PROMPT_HYPOTHESIS> in code block
Phase 3 (Enhanced Prompt Creation)
Applies prompt engineering best practices to create a measurably better version than the hypothesized original. The specific requirements (placeholders, numbered instructions, success criteria) ensure the new prompt follows modern prompt engineering standards rather than simply copying old patterns.
### Phase 3: Enhanced Prompt Creation
Create improved prompt (<NEW_PROMPT_v1>) that:
- Uses explicit <PLACEHOLDER> format for variables
- Contains numbered instruction sequences
- Includes success criteria rubric
- Specifies chain-of-thought privacy requirements
- Addresses gaps from Phase 1 analysis
- Uses modular, user-friendly structure
**Deliverable:** <NEW_PROMPT_v1> in code block
Phase 4 (Validation Testing)
Provides empirical evidence of prompt performance rather than theoretical assessment. Without actual testing, improvements remain hypothetical. The standardized output format enables systematic comparison in the next phase.
### Phase 4: Validation Testing
1. Generate 3 test outputs using <NEW_PROMPT_v1>
2. Format each as: <TEST_OUTPUT_1>[content]</TEST_OUTPUT_1>
3. Do not expose internal reasoning
Phase 5 (Systematic Evaluation)
Converts subjective quality assessment into objective, measurable criteria. The numeric rubric prevents bias and creates clear improvement targets, while the comparison table format makes deficiencies immediately visible and actionable.
### Phase 5: Systematic Evaluation
Rate each test output vs. original examples (1-10 scale):
| Criterion | TEST_1 | TEST_2 | TEST_3 | Mean |
|-----------------------------|--------|--------|--------|-------|
| Content Quality & Relevance | | | | |
| Structural Fidelity | | | | |
| Tone/Style Alignment | | | | |
| Completeness & Depth | | | | |
| Constraint Adherence | | | | |
| **OVERALL MEAN** | | | | |
Flag any category scoring lower than 7 for mandatory revision.
Phase 6 (Iterative Refinement)
Prevents the process from terminating with suboptimal results while avoiding endless iteration loops. The 3-iteration cap balances thoroughness with practical time constraints, and the score-based triggers ensure refinement only occurs when genuinely needed.
### Phase 6: Iterative Refinement
If any category scores lower than 8:
- Identify specific shortcomings
- Revise prompt (increment version: v2, v3)
- Re-test and re-evaluate
- **Maximum 3 iterations total**
Phase 7 (Final Deliverable)
Packages the results in immediately usable formats with sufficient context for implementation. The structured output ensures users receive both the refined prompt and the knowledge needed to apply it effectively, rather than just raw code without context.
### Phase 7: Final Deliverable
Provide:
1. <NEW_PROMPT_FINAL>: The final refined reverse-engineered prompt.
2. <MINIMUM_VIABLE_INPUT>: What the user input should include minimally to get good outputs.
3. <GUIDELINES>: Usage guidelines (implementation, variable insertion)
4. <SUMMARY>: Changes from hypothesized original. Performance enhancement rationale. Expected output characteristics
Quality Assurance
Defines explicit success criteria and edge-case handling to prevent common failure modes. Without these guidelines, the process could produce inconsistent results or fail when encountering unusual input conditions, undermining reliability in real-world usage.
## QUALITY ASSURANCE
### Success Definition
Process terminates when:
- All rubric categories score higher or equal to 8.0, OR
- 3 iterations completed (whichever first)
### Edge-Case Protocols
- **Contradictory examples:** Focus on dominant pattern, note discrepancies
- **Missing formatting:** Default to clear markdown structure
- **Multi-domain content:** Optimize for majority domain, flag others
- **Insufficient examples:** Request clarification or work with available data
Required User Input
Provides a clear template that reduces user confusion and ensures consistent input formatting. The placeholder structure prevents users from wondering where to insert their data, which is critical for a process that depends on precise input formatting.
If you are using this as part of an agent, you can put this in the user prompt and all the above instructions in the system prompt.
## USER INPUT
<MINIMUM_VIABLE_INPUT>
[Paste your input here]
</MINIMUM_VIABLE_INPUT>
<OUTPUT_1>
[Paste first example here]
</OUTPUT_1>
<OUTPUT_2>
[Paste second example here]
</OUTPUT_2>
<OUTPUT_3>
[Paste third example here]
</OUTPUT_3>
A Full Example To Run The Prompt
If you want to test it, to see if it fits your expectations, you can use the examples below. They demonstrate the expected input-output relationship with a concrete, realistic situation (the kind that I use often). Make sure you scroll down because it’s a long example.
<MINIMUM_VIABLE_INPUT>
TechFlow Solutions is a B2B SaaS startup offering workflow automation tools for mid-market companies. We've been in market for 8 months with 150 customers paying $200-800/month. Our core product automates repetitive business processes across departments like HR, finance, and operations. Current customers include manufacturing companies, professional services firms, and mid-sized retailers. We're seeing varied adoption patterns - some use us for simple task automation while others implement complex multi-department workflows. Our sales cycle ranges from 2 weeks to 4 months depending on company size and implementation complexity. We're preparing for Series A funding and need to refine our market segmentation strategy to optimize sales and marketing spend. We have basic firmographic data but lack deep behavioral insights about how different customer types use our platform.
</MINIMUM_VIABLE_INPUT>
<OUTPUT_1>
# Firmographic Market Segmentation Strategy
## Executive Summary
Based on your current customer base and product capabilities, implementing a structured firmographic segmentation approach will optimize your go-to-market efficiency and resource allocation ahead of Series A.
## Primary Segmentation Framework
### Segment 1: Mid-Market Manufacturers (Annual Revenue $10M-$100M)
**Key Characteristics:**
- 50-500 employees with established operational processes
- High process standardization needs across production and admin functions
- Longer implementation cycles (8-16 weeks) but higher contract values ($400-$800/month)
- Decision makers: Operations Directors, Plant Managers, IT Directors
**Value Proposition Focus:**
- Operational efficiency and cost reduction messaging
- Integration capabilities with existing ERP/MES systems
- Compliance and audit trail features
- ROI-focused case studies showing process time savings
### Segment 2: Professional Services Firms (Annual Revenue $2M-$50M)
**Key Characteristics:**
- 20-200 employees with project-based workflows
- High variability in process requirements across client engagements
- Medium implementation complexity (4-8 weeks) with moderate pricing ($200-$500/month)
- Decision makers: Managing Partners, Operations Managers, Practice Directors
**Value Proposition Focus:**
- Client delivery excellence and project efficiency
- Flexible workflow configuration capabilities
- Time tracking and billable hour optimization
- Client satisfaction and retention improvement metrics
### Segment 3: Growth-Stage Retailers (Annual Revenue $5M-$75M)
**Key Characteristics:**
- 25-300 employees with seasonal demand fluctuations
- Process automation needs across inventory, customer service, and back-office
- Rapid implementation preference (2-6 weeks) with budget-conscious approach ($200-$400/month)
- Decision makers: COOs, Store Operations Managers, IT Managers
**Value Proposition Focus:**
- Scalability during peak seasons
- Customer experience enhancement
- Inventory and supply chain optimization
- Multi-location standardization capabilities
## Implementation Roadmap
### Phase 1: Data Collection & Validation (2-4 weeks)
- Enhance customer data collection with firmographic details
- Conduct 15-20 customer interviews per segment
- Validate segment assumptions with usage analytics
- Create detailed buyer personas for each segment
### Phase 2: Go-to-Market Alignment (4-6 weeks)
- Develop segment-specific sales collateral and case studies
- Train sales team on segment identification and qualification
- Create targeted marketing campaigns for each segment
- Establish segment-specific success metrics and KPIs
### Phase 3: Optimization & Scaling (ongoing)
- Monitor segment performance and conversion rates
- Refine messaging based on market feedback
- Identify sub-segments within primary categories
- Scale successful approaches across similar market segments
## Success Metrics
- Segment-specific conversion rates from trial to paid
- Average contract value by segment
- Sales cycle length reduction
- Customer lifetime value by segment
- Market penetration rates within each segment
This firmographic approach provides clear targeting criteria while maintaining flexibility to adapt as your business scales toward Series A funding.
</OUTPUT_1>
<OUTPUT_2>
# Behavioral Market Segmentation Analysis
## Strategic Overview
Your diverse usage patterns indicate distinct behavioral segments that require differentiated engagement strategies. This behavioral segmentation will complement firmographic data to create a comprehensive targeting approach.
## Behavioral Segment Classification
### Segment A: Process Optimizers
**Usage Characteristics:**
- Implement 3-8 automated workflows within first 90 days
- High feature adoption rate (70%+ of available features used)
- Regular workflow modifications and improvements
- Strong integration usage with existing business systems
**Behavioral Indicators:**
- Frequent login patterns (daily/multiple times per day)
- Advanced feature utilization including conditional logic and API integrations
- Active participation in training sessions and support requests
- High engagement with educational content and best practices
**Engagement Strategy:**
- Position as workflow optimization partners, not just software vendors
- Provide advanced training and certification programs
- Offer strategic consultation on process improvement
- Create user community forums for peer learning and best practice sharing
**Revenue Characteristics:**
- Average contract value: $450-$800/month
- Expansion revenue potential: High (85% show usage growth over 6 months)
- Churn risk: Low (12% annual churn rate)
### Segment B: Task Automators
**Usage Characteristics:**
- Focus on 1-3 specific, repetitive task automations
- Limited exploration of advanced features
- Steady, consistent usage patterns
- Minimal customization or workflow modification
**Behavioral Indicators:**
- Moderate login frequency (2-3 times per week)
- Basic feature utilization focused on core automation capabilities
- Lower support ticket volume, mostly implementation-related
- Preference for self-service resources and documentation
**Engagement Strategy:**
- Emphasize ease of use and quick wins in messaging
- Provide templates and pre-built automation workflows
- Focus on time-saving and efficiency benefits
- Offer guided onboarding with clear milestone achievements
**Revenue Characteristics:**
- Average contract value: $200-$400/month
- Expansion revenue potential: Moderate (35% show usage growth)
- Churn risk: Medium (22% annual churn rate)
### Segment C: Experimental Users
**Usage Characteristics:**
- Inconsistent usage patterns with periods of high and low activity
- Wide feature exploration but shallow implementation depth
- Multiple workflow starts but fewer completions
- High support interaction during active periods
**Behavioral Indicators:**
- Sporadic login patterns with clustering around business cycles
- Feature trial behavior without deep implementation
- Higher support ticket volume relative to usage
- Mixed engagement with educational content
**Engagement Strategy:**
- Implement success milestone tracking and celebration
- Provide dedicated customer success management for guidance
- Create step-by-step implementation playbooks
- Offer regular check-ins and strategic planning sessions
**Revenue Characteristics:**
- Average contract value: $200-$350/month
- Expansion revenue potential: Variable (45% either expand significantly or churn)
- Churn risk: High (35% annual churn rate)
## Cross-Segment Insights
### Migration Patterns
- 25% of Task Automators evolve into Process Optimizers within 12 months
- Experimental Users require 4-6 months to stabilize into consistent usage patterns
- Process Optimizers rarely regress to simpler usage models
### Product Development Implications
- Process Optimizers drive feature complexity and integration requirements
- Task Automators validate core product-market fit and ease of use
- Experimental Users reveal onboarding friction points and support gaps
## Implementation Strategy
### Immediate Actions (0-30 days)
- Tag existing customers by behavioral segment using usage analytics
- Create segment-specific onboarding flows and success metrics
- Develop behavioral triggers for proactive customer success intervention
### Medium-term Initiatives (30-90 days)
- Build predictive models to identify segment classification early in customer lifecycle
- Create segment-specific retention and expansion campaigns
- Establish specialized support tracks for each behavioral type
This behavioral segmentation framework enables personalized customer experiences while identifying expansion opportunities and churn prevention strategies critical for sustainable growth.
</OUTPUT_2>
<OUTPUT_3>
# Needs-Based Market Segmentation Framework
## Strategic Foundation
Understanding your customers' underlying business needs, rather than just their demographics or behaviors, enables more precise value proposition development and solution positioning for optimal market penetration.
## Core Needs-Based Segments
### Segment 1: Compliance & Risk Management Focused
**Primary Business Need:**
Ensuring regulatory compliance, audit readiness, and risk mitigation across operational processes.
**Customer Profile:**
- Industries: Healthcare services, financial services, manufacturing with regulatory oversight
- Pain Points: Manual compliance tracking, audit preparation inefficiencies, regulatory reporting complexity
- Success Metrics: Audit pass rates, compliance cost reduction, risk incident prevention
**Solution Positioning:**
- Automated compliance workflow templates
- Comprehensive audit trails and documentation
- Real-time compliance monitoring and alerting
- Integration with regulatory reporting systems
**Value Proposition:**
"Transform compliance from a cost center to a competitive advantage with automated regulatory workflows that ensure 100% audit readiness while reducing compliance overhead by 60%."
**Sales Approach:**
- Lead with risk mitigation and compliance efficiency messaging
- Provide regulatory requirement mapping and gap analysis
- Offer compliance workflow audits and optimization services
- Position as regulatory technology partner, not just automation tool
### Segment 2: Operational Efficiency & Cost Reduction Focused
**Primary Business Need:**
Maximizing operational efficiency and reducing labor costs through process automation and optimization.
**Customer Profile:**
- Industries: Manufacturing, logistics, back-office heavy service providers
- Pain Points: High operational costs, manual process inefficiencies, resource allocation challenges
- Success Metrics: Process time reduction, labor cost savings, throughput improvements
**Solution Positioning:**
- Process efficiency analysis and optimization capabilities
- Resource allocation and workload balancing features
- Cost tracking and ROI measurement tools
- Integration with operational systems and data sources
**Value Proposition:**
"Achieve 40% operational efficiency gains and reduce process costs by $50,000+ annually through intelligent workflow automation that eliminates manual bottlenecks."
**Sales Approach:**
- Lead with ROI calculations and efficiency improvement metrics
- Provide operational assessment and process optimization consulting
- Offer pilot programs with measurable efficiency benchmarks
- Position as operational transformation partner
### Segment 3: Growth Scalability & Standardization Focused
**Primary Business Need:**
Building scalable, standardized processes that support rapid business growth without proportional increases in operational complexity.
**Customer Profile:**
- Industries: High-growth professional services, expanding retail chains, scaling technology companies
- Pain Points: Process inconsistency across locations/teams, scaling operational capabilities, maintaining quality during growth
- Success Metrics: Process standardization rates, scalability efficiency, quality consistency measures
**Solution Positioning:**
- Multi-location process standardization capabilities
- Scalable workflow templates and frameworks
- Performance monitoring and quality assurance features
- Change management and rollout support tools
**Value Proposition:**
"Scale your operations 3x faster while maintaining consistent quality and reducing management overhead through standardized, automated workflows that grow with your business."
**Sales Approach:**
- Lead with scalability and growth enablement messaging
- Provide growth readiness assessments and scalability planning
- Offer phased implementation aligned with expansion timelines
- Position as growth infrastructure partner
## Needs-Based Segmentation Implementation
### Customer Needs Assessment Framework
**Discovery Questions by Segment:**
*Compliance & Risk Management:*
- What regulatory requirements impact your operations?
- How do you currently manage audit preparation and documentation?
- What compliance-related incidents have you experienced?
*Operational Efficiency & Cost Reduction:*
- Where do you see the highest operational inefficiencies?
- What processes consume the most manual effort?
- How do you currently measure process costs and efficiency?
*Growth Scalability & Standardization:*
- What operational challenges do you anticipate with business growth?
- How consistent are your processes across different teams/locations?
- What prevents you from scaling operations more quickly?
### Segment-Specific Success Metrics
**Compliance & Risk Management:**
- Audit preparation time reduction: 70%+
- Compliance incident reduction: 90%+
- Regulatory reporting efficiency: 80%+
**Operational Efficiency & Cost Reduction:**
- Process time reduction: 40%+
- Labor cost savings: $50,000+ annually
- Operational error reduction: 85%+
**Growth Scalability & Standardization:**
- Process standardization achievement: 95%+
- Scaling efficiency improvement: 200%+
- Quality consistency maintenance: 98%+
## Go-to-Market Alignment
### Sales Team Specialization
- Assign account executives by needs-based segment expertise
- Develop segment-specific qualification frameworks and discovery methodologies
- Create specialized sales tools and ROI calculators for each segment
### Marketing Message Differentiation
- Develop distinct content strategies addressing each segment's core needs
- Create case studies and testimonials from each needs-based segment
- Implement targeted advertising campaigns with needs-specific messaging
This needs-based segmentation approach ensures your solution directly addresses customers' fundamental business drivers, creating stronger value proposition alignment and competitive differentiation essential for Series A growth trajectory.
</OUTPUT_3>
The Final Prompt
Steal this prompt, adapt it to your use cases, and run your own reverse-engineering experiments. Drop me a note on LinkedIn with your thoughts, ideas, or any twists you’ve added along the way.
(I guess it’d be annoying for you to have to copy-paste all the above blocks into one prompt, so the full prompt can be copied below (scroll down the code block).
Enjoy!
## ROLE
You are an expert prompt engineer specializing in reverse-prompt engineering, iterative optimization, and evaluation.
## GOAL
You will analyze output examples, reconstruct the original prompt that generated them, and create an improved version through systematic analysis and testing, using the workflow below.
## INPUT
You will receive three output examples (<OUTPUT_1>, <OUTPUT_2>, <OUTPUT_3>) generated by an unknown prompt and an input (MINIMUM_VIABLE_INPUT>) that is expected to generate those outputs.
## WORKFLOW
Follow the seven-phase protocol below. Think step by step. Keep detailed reasoning internal, expose only concise rationales and final deliverables. Never reveal full chain-of-thought processes.
### Phase 1: Deep Output Analysis
Analyze each example for:
- Content structure (headers, lists, paragraphs, formatting)
- Tone, style, and linguistic characteristics
- Format specifications and constraints
- Subject matter depth and technical level
- Implicit instructions evident in outputs
- Common themes across all examples
**Deliverable:** Analysis table with columns: *Feature*, *Evidence (≤20 words)*, *Applies to Example(s)*
### Phase 2: Prompt Reconstruction
Reconstruct the probable original prompt including:
- Role declarations and persona
- Context setup and background
- Input specifications
- Formatting directives
- Style requirements
- Constraints and limitations
**Deliverable:** `ORIGINAL_PROMPT_HYPOTHESIS` in code block
### Phase 3: Enhanced Prompt Creation
Create improved prompt (<NEW_PROMPT_v1>) that:
- Uses explicit <PLACEHOLDER> format for variables
- Contains numbered instruction sequences
- Includes success criteria rubric
- Specifies chain-of-thought privacy requirements
- Addresses gaps from Phase 1 analysis
- Uses modular, user-friendly structure
**Deliverable:** <NEW_PROMPT_v1> in code block
### Phase 4: Validation Testing
1. Generate 3 test outputs using <NEW_PROMPT_v1>
2. Format each as: <TEST_OUTPUT_1>[content]</TEST_OUTPUT_1>
3. Do not expose internal reasoning
### Phase 5: Systematic Evaluation
Rate each test output vs. original examples (1-10 scale):
| Criterion | TEST_1 | TEST_2 | TEST_3 | Mean |
|-----------------------------|--------|--------|--------|-------|
| Content Quality & Relevance | | | | |
| Structural Fidelity | | | | |
| Tone/Style Alignment | | | | |
| Completeness & Depth | | | | |
| Constraint Adherence | | | | |
| **OVERALL MEAN** | | | | |
Flag any category scoring lower than 7 for mandatory revision.
### Phase 6: Iterative Refinement
If any category scores lower than 8:
- Identify specific shortcomings
- Revise prompt (increment version: v2, v3)
- Re-test and re-evaluate
- **Maximum 3 iterations total**
### Phase 7: Final Deliverable
Provide:
1. <NEW_PROMPT_FINAL>: The final refined reverse-engineered prompt.
2. <MINIMUM_VIABLE_INPUT>: What the user input should include minimally to get good outputs.
3. <GUIDELINES>: Usage guidelines (implementation, variable insertion)
4. <SUMMARY>: Changes from hypothesized original. Performance enhancement rationale. Expected output characteristics
## QUALITY ASSURANCE
### Success Definition
Process terminates when:
- All rubric categories score higher or equal to 8.0, OR
- 3 iterations completed (whichever first)
### Edge-Case Protocols
- **Contradictory examples:** Focus on dominant pattern, note discrepancies
- **Missing formatting:** Default to clear markdown structure
- **Multi-domain content:** Optimize for majority domain, flag others
- **Insufficient examples:** Request clarification or work with available data
## USER INPUT
<MINIMUM_VIABLE_INPUT>
[Paste your input here]
</MINIMUM_VIABLE_INPUT>
<OUTPUT_1>
[Paste first example here]
</OUTPUT_1>
<OUTPUT_2>
[Paste second example here]
</OUTPUT_2>
<OUTPUT_3>
[Paste third example here]
</OUTPUT_3>