[Prompt Practical Tutorial] Use Gemini 2.5 Pro to handle seven major product documents

Yesterday we talked about how AI can solve the documentation problems of product managers. Today, we will teach you how to use Gemini 2.5 Pro. After all, no matter how good the tool is, it is useless if you don’t know how to use it.

Why choose Gemini 2.5 Pro?

I remember the first time I used Gemini 2.5 Pro to draw a business process diagram, it took me only 25 minutes to complete a task that would have taken me half a day. There are numerous AI tools on the market, but Gemini 2.5 Pro has three obvious advantages:

• • The context window reaches millions of tokens and can handle very long documents
• • Excellent multimodal capabilities, especially in chart generation
• • Strong logical reasoning ability, suitable for structured thinking

Core prompt word framework

No matter what type of document I'm working on, I use this basic framework:

You are a senior product expert with 10 years of experience in Internet products.
I am working on [document type] for the purpose of [specific goal].
The target user is [user description] and the main usage scenario is [scenario description].
Please help me complete [specific request].
Output format requirements: [format description].

This framework is not static, it needs to be added, deleted and adjusted according to specific needs.

Seven practical document prompts reference

1. Strategic planning tips

# Role
You are a professional product strategy planning expert, good at visual expression and SVG graphic design.

# Task
Generate a product strategy map in SVG format according to the following requirements, integrating the North Star Metric and Business Canvas elements.

# Format requirements
1. Structural level:
   - Top level: Corporate vision/mission
   - Middle level: North Star Indicator + Nine Elements of Business Canvas
   - Bottom layer: strategy implementation path

2. Visual standards:
   █ Color scheme: blue-gray main color (#2E86AB, #A7BED3) + highlight color (#F24236)
   █ Icon system:
     • Vision: Mountain Icon
     • Indicators: Dashboard icons
     • Business elements: corresponding field standard icons
   █ Layout logic: radial structure (vision center → indicator surround → business module outer circle)

3. Interaction Design:
   - Hover to display module detailed instructions
   - Click to expand implementation path details

# Key Elements
1. North Star Indicator Components:
   • Core indicator formula
   • Current value/target value comparison
   • Growth lever diagram

2. Business Canvas Module:
   Customer Segmentation → Value Proposition → Channels → Customer Relationships → Revenue Sources
   ↓
   Core resources → Key business → Important cooperation → Cost structure

3. Strategic Path:
   Product Roadmap (3-stage timeline)
   █ Resource requirements matrix
   █ Risk radar chart

# Quality Criteria
1. Information density: ≥ 0.8 effective information points per square centimeter
2. Visual hierarchy: Z-shaped reading flow (upper left → lower right)
3. Scalability: Reserve 20% blank area for iteration
4. Output specifications: SVG code must contain comment tags and responsive view boxes

# Behavioral Guidance
Please complete in three steps:
1. Confirm the integrity of strategic elements (verify the B=MAT model)
2. Constructing a visual semantic network (node ​​relationship ≥ 3 layers)
3. Implement noise reduction (remove decorative elements)

2. Product Evolution Roadmap Tips

# Role
You are a senior product strategy analyst, proficient in data visualization and product roadmap design

# Task
Design a dynamic and interactive product evolution roadmap, requiring:
1. Completely demonstrate the key milestones of the product from MVP to maturity
2. Integrate the three dimensions of technology evolution, user growth, and business value
3. Support multi-level information drilling (from annual goals → quarterly key results → monthly iterations)

# Chart Selection (Chart Selection Suggestions)
| Dimension | Recommended charts | Applicable scenarios | Recommended libraries |
|-------------|----------------------|---------------------------------|----------------|
| Timeline | Gantt chart | Display parallel development tasks across stages | Chart.js+Timeline plugin |
| Technology Evolution | Heatmap | Displaying the iteration density of technology stack | D3.js |
| User growth | Combination chart (Line+Bar) | Simultaneously display the number of users and retention rate | ECharts |
| Business Value | Radar | Multi-dimensional Indicator Comparison (Revenue/Profit Margin/Market Share) | Highcharts |

# Visualization Requirements
1. Time base:
   - X-axis: time scale (supports zooming from quarterly view → weekly view)
   - Mark key nodes: MVP release PMF verification Scale expansion

2. Interactive functions:
   - Hover to display iteration details (technical breakthroughs + user feedback)
   - Click to expand the associated resource input (human/funding)
   - Filter toggle business line view

3. Visual standards:
   █ Color matching principle: gradual color in stages (MVP light blue → mature dark blue)
   █ Icon semantics:
     ▲ Technological breakthrough
     ● User growth
     ★ Business Milestones
   █ Motion effect design: path growth animation (duration=800ms)

# Data Structuring (Data Structuring Suggestions)

javascript
{
  "phases": [
    {
      "name": "MVP Verification",
      "period": ["2023-Q2", "2023-Q4"],
      "techStack": ["Microkernel Architecture"],
      "metrics": {
        "users": 5000,
        "ltv": 120,
        "featureGroups": 3 
      },
      "dependencies": ["Cloud Service Procurement"]
    }
  ]
}

# Implementation Guidance
1. Development Priorities (Fogg Behavior Model):
   - Motivation: Display the routes of the last 6 months by default (reduces cognitive load)
   - Ability: Provides the function of exporting graphics to PNG/PDF
   - Trigger: Set up key node reminder function

2. Quality inspection checklist:
   - [ ] All time intervals are closed and have no overlap
   - [ ] Number of patents/papers annotated with technology evolution
   - [ ] User growth curve matches conversion funnel
   - [ ] Commercial value data unit is unified (10,000 yuan/USD)

3. Optimization suggestions:
   • Add probability annotation to technical risk nodes (50%↑ highlighted in red)
   • User growth index increased month-on-month (△5%)
   • Added competitor comparison layer (need to switch display)

3. User Story Map Tips

# Role
You are an agile coach who is proficient in user experience design and good at presenting user stories in a visual way.

# Task
To create a user story map with a hand-drawn style, you need to integrate the following elements:
1. Vertical stratification: User journey stages (discovery → decision → use)
2. Horizontal expansion: User activities → User tasks → User stories
3. Visual decoration: simulate whiteboard hand-painted effect + sticky note texture

# Hand-drawn Style Specification

vegalite
{
  "mark": {
    "type": "rect",
    "fill": "#FFF8DC", // kraft paper background color
    "stroke": "#5F4B32", // brown stroke
    "strokeWidth": 1.5,
    "strokeDash": [3,2], // dashed line effect
    "cornerRadius": 2
  },
  "background": "#F5F5F5", // whiteboard background color
  "config": {
    "text": {
      "font": "Comic Sans MS", // handwriting font
      "fontWeight": "bold"
    }
  }
}

# Core component design
1. User Journey Swimlanes:
   - Use wavy lines to separate stages (︿︿︿︿)
   - Stage title: Imitation highlighter effect (yellow semi-transparent rectangle)

2. Story Cards:
   █ Size: 120x80px (simulated sticky note)
   █ Style:
     • Default state: slight tilt (rotate: -2deg)
     • Hover state: darker shadow + sticky note curling effect
   █ Color coding:
     ? Functional requirements
     ? Technical Debt
     ? Innovative experiments

3. Associated elements:
   - Use pencil sketch lines to connect related stories (stroke-dasharray: 5,3)
   - Key points: hand-drawn star ⭐ and explosion sticker ? effects

# Interaction Requirements
1. Dynamic effects:
   - Added "tear off sticky note" animation during story
   - Drag and drop magnetic effect when sorting priorities

2. Annotation function:
   - Simulate whiteboard marker writing (retain pen pressure)
   - Support handwriting input conversion (circle → checkbox)

# Output configuration
1. Format options:
   - SVG (preserves vector hand-drawn details)
   - PNG (with paper texture background)

2. Layout template:
   Mobile: vertical stacking (showing the main user flow)
   █ Desktop: Horizontal expansion (full journey view)

# Quality checkpoints
1. Hand-painted authenticity:
   - [ ] There are moderate irregular edges
   - [ ] Text baseline slightly fluctuates
   - [ ] Color blocks have uneven coloring effect

2. Story integrity:
   - [ ] Each user activity has ≥ 3 decomposition tasks
   - [ ] Key pain points are marked with a ? icon
   - [ ] Technical dependencies are marked with a ?link trigger

# Behavioral Design Tips
According to the Fogg Behavioral Model (B=MAT):
1. Motivation: Display MVP scope stories by default (reduces decision pressure)
2. Ability: Provides quick generation of story templates (enter "as [character]...")
3. Trigger: Set the completion animation (cross-out effect + cheering sound effect)

4. Product Function Architecture Diagram Prompt Words

# Role
You are a senior product architect and industry analysis expert.

# Task
1. **Analysis and understanding**: Based on the product’s core value, target users, and key features list I provided, deeply analyze its internal logical relationship.
2. **Build architecture**: Organize the functional modules according to the pyramid structure of "core layer-application layer-service layer" or "front-end-middle-back-end" to build a logically clear and scalable product functional architecture.
3. **Generate a graph**: Use Mermaid syntax to create a visual product functional architecture diagram (`graph TD`).
4. **Provide action suggestions**:
    * **Interpretation of the architecture**: Briefly explain the overall architecture design ideas and the core responsibilities of each module.
    * **Planning Path (B=MAT)**: Propose a phased, executable implementation roadmap (such as MVP -> Core Version -> Full Version), clarify the key functional points of each stage, and make it easy to start and execute.

# Format
Please strictly follow the following Markdown structure to ensure that the content is clearly structured and the key points are highlighted:

## 1. Product Function Architecture Diagram (Mermaid)

mermaid
graph TD
    Subgraph User End (Frontend)
        A[Core Function A] --> A1[Sub-Function A1]
        A[Core Function A] --> A2[Sub-Function A2]
    end
    Subgraph Management Backend
        B[Core Function B] --> B1[Sub-Function B1]
        B[Core Function B] --> B2[Sub-Function B2]
    end
    A --> B

## 2. Interpretation of the core architecture
**Overall Design**: [Here is a summary of the top-level design ideas of the architecture]
* **Module Responsibilities**:
    * **Module 1**: [Describe the core value and main functions of this module]
    * **Module 2**: [Describe the core value and main functions of this module]
    * ...

## 3. Phased Action Plan
**Phase 1 (MVP):
    * **Objective**: Validate core business model.
    * **Key functions**: [List the most core and basic functions]
**Phase 2 (Core Version)**:
    * **Goal**: Improve user experience and increase retention.
    **Key Features**: [List the features that enhance the core experience]
**Phase 3 (Full Version)**:
    * **Goal**: Expand business boundaries and build an ecosystem.
    * **Key functions**: [List the functional points for business expansion or ecosystem construction]

---
## Product Information
* **Product Name**: []
* **Core Values**: []
* **Target users**: []
* **Features**:
    * Function 1
    * Function 2
    * …

5. Lean Value Tree Tips

# Role
You are a senior Lean Enterprise Coach and product strategy expert, proficient in connecting corporate vision with the daily development activities of front-line teams.

# Task
1. **Analyze and Build**: Analyze and build a structured Lean Value Tree based on the “Product Vision”, “Strategic Goals” and “Initial Initiatives” I provided.
2. **Clear hierarchy**: Organize information into a clear hierarchy of "Vision -> Strategic Goals -> Product Outcomes -> Functional Initiatives (Outputs)" to ensure that top-level goals can be effectively broken down.
3. **Refine hypothesis**: For each "functional measure", extract the "product outcome" hypothesis behind it. In other words, clarify "what we think will cause changes in user behavior when we do something".
4. **Visualization**: Use Mermaid syntax (`graph TD`) to create a visual Lean Value Tree chart.
5. **Drive Action (B=MAT)**:
    * **Identify key assumptions**: Point out the core assumptions in the entire value tree that have the highest risk and need to be verified first.
    **Design a minimal experiment**: For the core hypothesis, design a specific, low-cost, and easy-to-execute verification step (for example: make a prototype, conduct an interview, launch an A/B test) to make the verification behavior simple and feasible.

# Format
Please strictly follow the following Markdown structure to ensure logical clarity and operability:

## 1. Lean Value Tree Diagram (Mermaid)

mermaid
graph TD
    subgraph Vision
        A("Achieve personalized health management")
    end

    Subgraph Goals [Strategic Goals]
        B("Increase user activity")
        C("Verify payment mode")
    end

    Subgraph Outcomes
        D("Users record health data more frequently")
        E("Users are willing to pay for advanced analysis reports")
    end

    Subgraph Outputs [Functional Measures]
        F("Develop a gamified data recording module")
        G("Introducing AI in-depth health analysis report")
    end

    A --> B;
    A --> C;
    B --> D;
    C --> E;
    D --> F;
    E --> G;


## 2. Core Value Assumptions
* **Hypothesis 1**: We believe that through **[Functional Initiative F: Develop a gamified data recording module]**, **[Product Outcome D: Users record health data more frequently]** will be achieved, thereby promoting **[Strategic Goal B: Improve user activity]**.
* **Hypothesis 2**: We believe that through **[Functional Initiative G: Launch AI Deep Health Analysis Report]**, we will achieve **[Product Outcome E: Users are willing to pay for advanced analysis reports]**, thereby promoting the achievement of **[Strategic Goal C: Verify the payment model]**.
* ...

## 3. Action and Verification Plan
* **Identification of key assumptions**: The core assumption with the highest risk and the most uncertainty at present is **[e.g.: Assumption 2]** because it is directly related to the product's business closed loop.
* **Minimize verification steps**:
    1. **Trigger point (Motivation)**: Send an email to 100 highly active users to introduce the concept of “AI Deep Health Analysis Report”.
    2. **Simplify Ability**: Place a pre-order button in the email that says "I'm interested, notify me when it's online", and after clicking it, you only need to enter your email address.
    3. **Prompt**: Track click-through rate and pre-order conversion rate as initial signals to verify user willingness to pay.

---
## Product Information
* **Product Vision**: [For example: To become the world's leading online collaborative whiteboard tool]
* **Strategic Goals**: [For example: 1. Improve the team collaboration efficiency of enterprise users. 2. Expand the education market share.]
* **Initiative Ideas**: [For example: 1. Add real-time audio and video call function. 2. Develop a template library suitable for tablet teaching. 3. Introduce AI assistant to automatically organize notes.]

6. Tips for in-depth research on enterprises (Source: Mr. Yao Jingang)

# Enterprise in-depth research and analysis expert

You are a senior enterprise strategy analysis expert with more than 15 years of consulting experience. You are good at using tools such as Google Deep Research to conduct in-depth research on enterprises and have keen business insight and systematic thinking ability. You have served many Fortune 500 companies and have rich practical experience in digital transformation, business model innovation, strategic reconstruction and other fields.

## Core Competencies
- **Multi-dimensional analysis**: comprehensive analysis of finance, market, technology, operations and strategy
- **Data-driven**: Objective analysis based on reliable data sources
- **Forward-looking insights**: Identify changing trends and potential opportunities
- **Practical orientation**: Provide executable strategic suggestions

# Task
Conduct in-depth research and analysis on target companies and output structured corporate insight reports.

## Analysis Methodology
- **SWOT analysis**: systematically evaluate the internal and external environment
- **Porter's Five Forces Model**: In-depth analysis of the competitive landscape
- **Value chain analysis**: Identify the core value creation links
- **Business Canvas**: Sorting out the logic of business models
- **Scenario analysis**: predicting various development possibilities
- **Benchmark analysis**: Learn from the best practices of the same industry

## Data Collection Standards
### First-hand data (priority: high)
- Financial statements, annual reports, quarterly reports
- Official announcements, investor relations materials
- Management interviews, performance briefings
- Patent applications, technical white papers

### Secondary data (priority: medium)
- Authoritative industry reports (McKinsey, BCG, Deloitte, etc.)
- Professional media reports (36Kr, Titanium Media, etc.)
- Analyst research reports (brokerage firms, investment banks)
- User reviews and social media buzz

### Data quality requirements
- **Timeliness**: Priority is given to using data from the past two years, and key indicators require a five-year trend
- **Credibility**: Label the source of the data and distinguish between confirmed facts and inferred judgments
- **Completeness**: Multi-source verification to fill in information gaps

## Core analysis dimensions

### 1. Basic information portrait
- Company profile: founding date, registered capital, equity structure
- Development history: key milestones and major turning points
- Organizational structure: management background, governance structure
- Core team: founder's resume, senior management team capabilities

### 2. In-depth analysis of business
- **Business model analysis**: value proposition, revenue model, cost structure
- **Product service matrix**: core products, product life cycle, innovation pipeline
- **Customer value proposition**: target customer groups, demand pain points, solutions
- **Revenue structure**: business line contribution, geographical distribution, customer concentration

### 3. Market environment assessment
- **Industry landscape analysis**: market size, growth trend, development stage
- **Competitive landscape**: major competitors, market share, differentiated positioning
- **Industry chain position**: upstream and downstream relationships, bargaining power, and substitution threats
- **Policy environment**: regulatory policies, industry standards, government support

### 4. Financial Health Diagnosis
- **Key financial indicators**: revenue, profit, cash flow, ROE/ROA
- **Profitability analysis**: gross profit margin, net profit margin, profit quality
- **Capital efficiency assessment**: asset turnover, leverage level, capital structure
- **Growth analysis**: revenue growth rate, profit growth rate, sustainability

### 5. Review of innovation capabilities
- **Technical R&D strength**: R&D investment, number of patents, technical team
- **Transformation of innovative achievements**: New product contribution, technology commercialization capabilities
- **Future technology layout**: cutting-edge technology reserves, strategic technology directions
- **Innovation Ecosystem**: Industry-University-Research Cooperation, Open Innovation Platform

### 6. Strategic resource assessment
- **Core assets**: brand value, technological barriers, data assets
- **Partners**: strategic alliances, supplier relationships, channel networks
- **Human Resources**: Talent structure, core team stability, corporate culture
- **Intangible assets**: intellectual property, goodwill, customer relationships

## Breakout Leverage Identification Framework

### Growth Levers
- **Market penetration**: Existing market share can be increased
- **Market expansion**: new regions, new customer groups, new scenarios
- **Product innovation**: product line extension, function upgrade, cross-border integration
- **Channel optimization**: online and offline integration, channel sinking, direct sales model

### Efficiency levers
- **Digital transformation**: process automation, data-driven decision making
- **Supply chain optimization**: cost control, response speed, flexible manufacturing
- **Organization optimization**: flat management, agile organization, and improved human efficiency
- **Resource allocation**: Focus on core businesses and divest non-core businesses

### Innovation Leverage
- **Technological breakthrough**: core technology upgrade, new technology application
- **Model innovation**: business model reconstruction, value chain reorganization
- **Eco-cooperation**: Platform strategy, ecosystem construction
- **Cross-border integration**: Blurred industry boundaries and emerging business formats

## Risk Assessment Matrix

| Risk Type| Probability Assessment| Impact Level| Response Strategy| Monitoring Indicators|
|---------|----------|----------|----------|----------|
| Market Risk | High/Medium/Low | High/Medium/Low | Specific Measures | Key Indicators |
| Technical Risk | High/Medium/Low | High/Medium/Low | Specific Measures | Key Indicators |
| Financial Risk | High/Medium/Low | High/Medium/Low | Specific Measures | Key Indicators |
| Operational Risk | High/Medium/Low | High/Medium/Low | Specific Measures | Key Indicators |
| Policy Risk | High/Medium/Low | High/Medium/Low | Specific Measures | Key Indicators |

# Output format

## Analysis process description
During the analysis, I will:
1. **Hypothesis confirmation**: First put forward 3-5 key hypotheses, please confirm or modify them
2. **Information Supplement**: If the information is insufficient, we will actively ask for supplementation
3. **Stage confirmation**: Pause at key conclusions to ask for your opinions and supplements
4. **Preview review**: Before the final report, an executive summary will be provided for your review

## Report structure template

{
# [Company Name] In-depth Research and Analysis Report

## Executive Summary
- Core findings (3-5 key insights)
- Main recommendations (2-3 strategic directions)

## 1. Panoramic picture of the company
### 1.1 Basic Information
### 1.2 Development trajectory
### 1.3 Core Competitiveness

## 2. In-depth analysis of business
### 2.1 Business Model Analysis
### 2.2 Product Service Matrix
### 2.3 Customer Value Proposition

## 3. Market position assessment
### 3.1 Industry landscape analysis
### 3.2 Competitive Advantage Assessment
### 3.3 Market share trends

## 4. Financial Health Diagnosis
### 4.1 Key Financial Indicators
### 4.2 Profitability Analysis
### 4.3 Capital Efficiency Assessment

## 5. ABILITY
### 5.1 Technical R&D Strength
### 5.2 Transformation of Innovation Achievements
### 5.3 Future Technology Layout

## 6. Identify risks and opportunities
### 6.1 Main risk factors
### 6.2 Market Opportunity Analysis
### 6.3 Response strategy suggestions

## 7. Breakout Leverage Discovery
### 7.1 New Growth Engine
### 7.2 Efficiency Improvement Points
### 7.3 Model Innovation Opportunities

## 8. Strategic Proposal
### 8.1 Short-term action plan (6-12 months)
### 8.2 Mid-term development path (1-3 years)
### 8.3 Long-term vision planning (3-5 years)

## Appendix
- Data source description
- Analytical Methodology
- Important assumptions
}

## Output requirements
   * Data support: All opinions must be supported by specific data or cases
   * Logically rigorous: the analysis process follows cause and effect, and the conclusions are traceable
   * Action-oriented: Suggestions are specific and executable, including implementation steps
   * Risk Warning: Clearly point out the limitations and uncertainties of the analysis
   
Target Company: {{Company Name}}
Industry: {{Industry}}
Analysis focus: {{Specific focus}}

7. Business process diagram prompt words

# Role
You are a senior business analyst and process optimization consultant, good at organizing complex business operations into clear views and accurately locating efficiency bottlenecks.

# Task
1. **Analysis and sorting**: Based on the "business process description" I provided, identify all the "participating roles/systems", "core activities", "decision nodes" and "task flows".
2. **Build a flow chart**: Use the Mermaid syntax's swimlane diagram function to create a visual business process diagram. The responsibilities of each role/department and the flow of tasks between them (handoffs) must be clearly displayed.
3. **Interpret the process**: Use concise language to describe the core steps in the diagram in stages.
4. **Make optimization suggestions (B=MAT)**:
    **Identify Pain Points**: Review the entire process and identify key nodes that may cause delays, increase costs or operational cumbersomeness (i.e. increase user friction).
    * **Provide simplified solutions (Actionable Steps)**: For each identified bottleneck, a specific, actionable optimization suggestion is proposed (such as: automating a step, merging two links, providing a template, etc.) to make the process simpler and easier to execute.

# Format
Please strictly follow the following Markdown structure to ensure that the content is professional, clear and highly operational:

## 1. Business process diagram (Mermaid)

mermaid
graph TD
    Subgraph User
        A[Initiate a refund application] --> B{Are the application materials complete?};
    end

    Subgraph Customer Service
        B -- Yes --> C[Review Application];
        C --> D{Is it in compliance with the refund policy?};
        D -- No --> F [Reject the application and state the reasons];
    end

    Subgraph Finance Department
        D -- Yes --> E[Execute refund operation];
        E --> G[Notify the user of successful refund];
    end

    Subgraph System
        F --> H[Send rejection notice];
        G --> I[Send Success Notification];
    end
    
Business process information
Process name: []
Core objectives: []
Participating roles/departments: []
Description of process steps:
[Please describe the process in detail in order. For example:
1. …. 
2. …. 
3. …
4. …
5. … (please continue to add)]

Solutions to common problems

Problem 1: Output is too general Solution: Be specific. "Please list 5 key quantifiable indicators"

Problem 2: Not in line with business reality Solution: Provide more context. "Considering that we are a local SME, please adjust the suggestions"

Problem 3: The format does not meet the requirements Solution: Clearly demonstrate. "Please follow the title level and style of the example below"

Three tips to improve results

1. Interaction in stages

Don't expect to get perfect results with just one tip. My experience is to use three rounds:

1. 1. Round 1: Generate the overall framework
2. 2. Round 2: Fill in key details
3. 3. The third round: Optimize the expression

2. Provide examples

Saying "like this..." to Gemini is more effective than a long description. You can attach:

• • Similar documentary clips that you appreciate
• • Preferred chart style
• • Reference Case Links

3. Set constraints

Clear constraints often inspire better output, such as: "Summarize the core strategy in no more than 300 words" "Use no more than three colors in the chart" "Avoid using technical terms"

[Some small advanced ideas]

Prompts are not fixed formulas. It is easy to find that after collecting hundreds of so-called "perfect prompts", the results are mixed. What is really important is to understand the thinking behind the prompts so that you can know how to fine-tune them.

Why are your prompts always not effective enough?

Some people complained: "The product architecture diagrams generated by Gemini are always short of details." After reading the prompts, they found the problem - it was full of technical terms, but it didn't make it clear what problems the product was going to solve.

It's like if you go to a restaurant and just tell the waiter "I want one dish", the result will naturally be unsatisfactory. A good way to order is: "I want something light, for two people, I don't like onions, ginger and garlic, and my budget is about 100 yuan."

The same is true for good prompt words, which require three key elements:

• • Clarify the intention (what problem do you want to solve)
• • Provide context (what is happening now)
• • Setting boundaries (what not to do)

Dynamically adjust the four dimensions of prompt words

1. Target granularity adjustment

Last week, I was doing a strategic plan for overseas markets, and I tried three different granularity prompts:

• • Beginner's Edition: "Developing an Overseas Market Strategy"
• • Intermediate version: "Developing a three-year strategy for the Southeast Asian market for Chinese SaaS companies"
• • Advanced version: "Develop a market expansion strategy for Company A (a leading HR SaaS company in China) in Malaysia, focusing on solving localization adaptation issues"

The results are obviously different: the entry-level version talks about it in general terms, while the advanced version directly provides specific solutions including local payment integration, multi-language support, etc.

2. The art of role definition

I found that many people overlook the importance of role setting. Compare these two:

• • Regular version: "Generate a product roadmap"
• • Advanced version: "You are a CPO who has managed three products with tens of millions of users. Now..."

The latter’s output is obviously more strategic, because AI really simulates the way senior product managers think.

3. The magic of constraints

Unconstrained prompt words are like a desert without boundaries. Effective constraints include:

• • Format requirements: "Presented in Markdown table with timeline"
• • Content restrictions: "Do not discuss technical implementation details"
• • Style guide: "Use non-technical language that product managers can understand"

Here’s a good idea: let AI constrain itself. “If you were a serious consultant, what three questions would you raise about this solution?”

4. Injection of thought chain

The most easily overlooked thing is the guidance of the thought process. Compare these two:

• • Basic version: "Analyze competitors"
• • Advanced version: "Analyze according to the steps of ①determine the scope of competitive products → ②collect public data → ③analyze the product matrix → ④evaluate technical strength → ⑤summarize differentiation opportunities"

The latter allows AI to follow your thinking path and produce more organized output

Five key points to learn from masters

After reading many tips from AI experts, I summarized their common points:

1. Teach thinking methods first

Just like teaching new people to make products, instead of giving them templates directly, you explain the methodology behind it. For example: "When making a user story map, you should first sort out the complete user journey, then break down the key nodes, and finally fill in the specific stories."

2. Provide an analytical framework

Giving AI a clear thinking framework can greatly improve quality. For example: "Use Porter's Five Forces Model to analyze industry competition" "Use SWOT method to evaluate strategic options"

3. Demonstrate high-quality output

It is often most effective to present a sample of what you expect: "See the structure and depth of the following case study: [insert excellent example]"

4. Create a feedback loop

Let AI continue to optimize: "Based on the feedback from the last review, focus on improving these three aspects..."

5. Set up authentication mechanism

Ask AI to self-examine: "Generate five criteria to evaluate the professionalism of this research report"

Practical case: Intelligent process of dynamic adjustment

When I was recently leading a team to plan an intelligent customer service product, we iterated the prompt words like this:

Initial prompt: "Generate a product feature list for smart customer service"

First round of optimization: "As the product director of a leading cloud communication platform, I planned the functional architecture of an intelligent customer service system for medium and large enterprises. I focused on integrating with existing CRMs to avoid homogenization with competing products."

Second round of enhancements: "Based on the architecture design of 1) basic capability layer 2) intelligent interaction layer 3) data analysis layer 4) ecological integration layer. Refer to the differentiated highlights of Zendesk and Freshdesk to highlight our advantages in multimedia support and knowledge graphs"

Final version: "Assume that we have just received $25 million in Series B funding and need to present a three-year plan for our intelligent customer service product to the board of directors. Please create a complete framework that includes technical roadmaps, functional evolution, and business goals. Special requirements: There must be quantifiable milestones every quarter, and the risk section must cover policy compliance and ethical issues."

Each round of adjustments makes the output more precise and in-depth

Three advanced techniques worth collecting

1. Meta-hint lexicon

Here’s a trick for you: Use AI to optimize the prompt words themselves: “Please analyze the problems with the following prompt words and make three suggestions for improvement: [your prompt words]”

2. Mirroring

Let AI simulate your thinking: "How would I analyze this problem? Please answer according to my thinking habits and expression style."

3. Expert Committee Law

Inspire a more comprehensive perspective: "Assume that you are a committee consisting of a product director, a technical expert, and a user experience designer. Please evaluate this solution from your own perspective."

AI is like an intern - the more vague the task you give it, the worse the work it will turn in. Clear guidance will make it more valuable