Day 3: Five Questions That Build Better Prompts

From:

Prompt Guy <thinkaiprompt@mail.beehiiv.com>

To:

Hidden Recipient <hidden@emailshot.io>

Date:

2/28/2026, 1:56 PM(11 days ago)

Hey There,

Here's What's Different Today

Days 1-2: You learned foundations. You used working prompts.

Today: You become the one building the prompts.

Not copying templates. Actually thinking through what your prompt needs—and building it from scratch.

The shift: Instead of asking "how do I use this prompt?" you'll ask "what does my prompt need?" Then you'll build it. That's the difference between following instructions and understanding the system.

📚 What You're Learning Today

Four major things:

What nesting actually is. The real explanation. When you need it.
The Five Questions to think through BEFORE you build. How to prepare your thinking.
What to put IN the brackets: The thinking process. How to know what belongs where.
How to arrange your prompt from start to finish: The exact order. The layers. When you're done.

By the end of this email, you'll be able to take any content task and build a working prompt—from scratch.

Part 1: Understanding Nesting (What It Actually Is)

🎯 The Folder Analogy

You know what a folder inside a folder is on your computer, right?

You might have:

A folder called "Projects"
Inside that folder, a folder called "Blog Posts"
Inside THAT folder, individual blog post files

That's nesting.

Things organized inside other organized things.

In JSON prompting, nesting works the same way.

Instead of creating and organizing ONE thing, you create and organize a collection of organized things.

One prompt. Multiple organized pieces. All structured the same way. All inside one container.

🎬 The Visual Difference: Non-Nested vs. Nested

ONE blog post (no nesting needed):

{   

"task": "blog_post",  
 "title": "How to Use AI for Email Marketing",   
"audience": "Small business owners",   
"sections": [     { "section_title": "Benefit 1", 
"content": "..." },    
 { "section_title": "Benefit 2", "content": "..." }   ]


 }

Simple. One post. Done.

THREE blog posts with same structure (this needs nesting):

{   


"task": "batch_blog_creation",   
"series_name": "AI Tools For Professionals",   
"blog_posts": [     

{      

 "post_number": 1,      
 "title": "AI Tools For Writers",      
 "audience": "Content creators",      
 "sections": [         { "section_title": "Benefit 1", 
"content": "..." },         { "section_title": "Benefit 2", "content": "..." }       ]     },   
 
 {     

  "post_number": 2,     
  "title": "AI Tools For Marketers",     
  "audience": "Growth professionals",       "sections": [         { "section_title": "Benefit 1", "content": "..." },       
  { "section_title": "Benefit 2", "content": "..." }       ]     },     
{      

 "post_number": 3,    
   "title": "AI Tools For Entrepreneurs",     
  "audience": "Business owners",       "sections": [         { "section_title": "Benefit 1", "content": "..." },         { "section_title": "Benefit 2", "content": "..." }       ]     }   ] 


}

See it? You put three complete, organized blog posts INSIDE the "blog_posts" array. Each post has the same structure. AI returns all three at once. All organized. One prompt. Done.

That's nesting.

🔍 Where Nesting Actually Happens In The Brackets

The "blog_posts": [ ] part? That's the container—the array that holds multiple things.

The { } INSIDE the [ ]? Those are each complete, organized structures nested inside.

The pattern looks like:

[ { complete structure 1 }, { complete structure 2 }, { complete structure 3 } ]

Each { } is a full organized thing. All the { } go inside [ ]. That's nesting.

❌ vs ✅ Why Nesting Matters

❌ Without Nesting

You: Write prompt for blog 1, run it, get output.

Then: Write prompt for blog 2, run it, get output.

Then: Write prompt for blog 3, run it, get output.

Total: Three separate prompts. More work.

✅ With Nesting

You: Write ONE nested prompt for all three.

AI returns: All three organized together.

Total: One prompt. One time. Done.

🎯 When Do You Actually Need Nesting?

Ask yourself these three questions:

Question 1: Do I need multiple pieces of content?

If NO → Don't nest.

If YES → Go to Question 2.

Question 2: Do all pieces need the SAME structure?

If NO → Don't nest.

If YES → Go to Question 3.

Question 3: Do I want them organized together in ONE output?

If NO → Run separately.

If YES → USE NESTING.

Part 2: The Five Questions (Think Before You Build)

🧠 What Your Prompt Needs To Answer

Before you write anything, ask yourself these Five Questions:

The Five Questions:

1. WHAT are you asking for?

Be specific. "Blog post about email marketing" not "write something."

2. WHO is it for?

Define your audience. Small business owners? Students? This changes everything.

3. HOW should it sound?

The tone. Professional? Funny? Casual? What should it avoid?

4. WHAT should be included?

The requirements. Headings? Bullets? Examples? CTAs? What can't be missing?

5. HOW should it be organized?

The output format. Sections? Lists? Intro? Conclusion?

🎬 Real Example: Thinking Through A Blog Post

You want to write: "How to Use AI for Email Marketing"

WHAT?

A blog post about using AI tools for email marketing. How to write copy with AI, personalize campaigns, and automate workflows.

WHO?

Small business owners who manage their own marketing. No marketing team. They want practical tips today.

HOW?

Conversational. Like talking to a friend. Simple language. No jargon. Include real examples. Make it actionable.

WHAT TO INCLUDE?

Engaging intro. 3 main ways AI helps. Real examples for each. One tip per section. Strong conclusion with CTA.

HOW ORGANIZED?

Clear sections with headers. Bullet points for tips. Examples highlighted. CTA at the end. 1500-2000 words.

Now you know exactly what to build.

Part 3: What Goes IN The Brackets? (The Thinking Process)

🔑 How To Think About What Belongs In Your Brackets

Here's the secret: Everything in your brackets should be information that answers one of your Five Questions.

For each bracket section, ask: "What do I need to tell AI about this?"

For the TASK bracket: "What exactly am I asking for?" → Include task name, content type, topic.

For the AUDIENCE bracket: "Who is this for?"

→ Include who they are, what they care about, their experience level.

For the TONE bracket: "How should it sound?"

→ Include tone, voice, what to avoid, what to include.

For the FORMAT bracket: "How should it look?"

→ Include headings, bullets, sections, length.

For the REQUIREMENTS bracket: "What can't be missing?"

→ Include must-haves, structure, examples.

📝 Real Walkthrough: Building Brackets For A Blog Post

Let's build a blog post prompt step by step.

I'll show you exactly what goes in each bracket and WHY.

STEP 1: The TASK bracket — "What am I asking for?"

You ask yourself: What exactly do I need? A blog post. About what? AI for email marketing.

{  

 "task": "blog_post_writing",   

"topic": "How to Use AI for Email Marketing",   

"content_type": "educational blog post"

Why these fields? Because AI needs to know:

(1) You want a blog post written,

(2) The specific topic,

(3) What type of blog post.

STEP 2: The AUDIENCE bracket — "Who is this for?"

You ask yourself: Who will read this?

Small business owners. What are they like? Busy. Want practical tips. No team.

"target_audience": "Small business owners",  

 "audience_profile": {     "who_they_are": "People managing their own marketing",    

 "what_they_want": "Practical, actionable tips they can use today",     

"pain_point": "They're overwhelmed with marketing tasks"   },

Why these fields? Because AI needs to understand

(1) Who you're writing for,

(2) What they're like,

(3) What they need.

This changes the entire tone and content.

STEP 3: The TONE bracket — "How should it sound?"

You ask yourself: What voice do I want?

Conversational. What should it avoid? Jargon. What should it include? Real examples.

  "tone": "conversational, helpful, actionable",   "voice_requirements": {     "style": "Write like you're talking to a smart friend",     "avoid": ["corporate jargon", "overly technical terms", "vague advice"],    
 "include": ["real examples", "step-by-step instructions", "specific tool names"]   },

Why these fields? Because AI needs to know

(1) What tone to use,

(2) What to avoid,

(3) What to include.

This ensures it sounds like you want it to.

STEP 4: The FORMAT bracket — "How should it be organized?"

You ask yourself: What structure do I want?

Headings. Bullet points.

How long? 1500-2000 words.

Should it have examples? Yes.

  "output_format": {     "include_headings": true,     "include_subheadings": true,     
"use_bullet_points": true,    
 "include_real_examples": true,    
 "include_action_steps": true,    
 "word_count": "1500-2000 words"   },

Why these fields? Because AI needs to know exactly how to organize the information. What goes where. What format it should use.

STEP 5: The REQUIREMENTS bracket — "What can't be missing?"

You ask yourself: What absolutely must be in this post?

An intro. 3 main sections. Real examples.

Conclusion with CTA.

  "requirements": {     "word_count_range": "1500-2000 words",     "must_include": [       "Engaging introduction that hooks readers",       "3 main ways AI helps with email marketing",    

   "Real, specific examples for each way",       "One actionable tip per section",       

"Strong conclusion with clear CTA"     ]   } }

Why these fields? Because these are your non-negotiables.

The things that if they're missing, the post isn't what you wanted.

💡 The Bracket Thinking Pattern (Use This For Everything)

Here's how to think about what goes IN your brackets every time:

1. Ask yourself: "What question does AI need answered?"

Example: "What am I asking for?" or "Who is this for?" or "How should it sound?"

2. Answer that question with specific details.

Example: Not "Blog post" but "Blog post about AI for email marketing for small business owners"

3. Put that answer IN the bracket.

Example: "topic": "How to Use AI for Email Marketing", "target_audience": "Small business owners"

4. Ask the next question and repeat.

Keep going until you've answered all Five Questions.

Part 4: How To Arrange Your Prompt (The Five Layers)

🏗️ The Exact Order To Build In

Prompts have a natural, logical order.

Follow this:

LAYER 1: THE BASICS

What are you asking for? Task, content type, topic.

LAYER 2: THE AUDIENCE

Who is this for? Audience, experience level, their needs.

LAYER 3: THE VOICE

How should it sound? Tone, voice, what to avoid, what to include.

LAYER 4: THE OUTPUT

How should it be organized? Format, structure, length, sections.

LAYER 5: THE REQUIREMENTS

What can't be missing? Must-haves, specifics, non-negotiables.

📋 The Complete Blog Post Prompt (All Five Layers Together)

{   "task": "blog_post_writing",   "topic": "How to Use AI for Email Marketing",   "content_type": "educational blog post", 


     "target_audience": "Small business owners",   "audience_profile": {     "who_they_are": "People managing their own marketing",     "what_they_want": "Practical, actionable tips",     "pain_point": "Overwhelmed with marketing tasks"   },     


 "tone": "conversational, helpful, actionable",   "voice_requirements": {     "style": "Write like you're talking to a smart friend",     "avoid": ["corporate jargon", "overly technical terms"],     "include": ["real examples", "step-by-step instructions"]   },      


"output_format": {     "include_headings": true,     "include_subheadings": true,     "use_bullet_points": true,     "include_examples": true,     "word_count": "1500-2000 words"   },      


"requirements": {     "must_include": [       "Engaging introduction",       "3 main ways AI helps with email",       "Real, specific examples for each",       "One actionable tip per section",       "Strong conclusion with CTA"     ]   } }

Part 5: How To Know When You're Done

✅ The "Your Prompt Is Complete" Checklist

Your prompt is ready when you answer YES to ALL of these:

✓ Task is crystal clear

Could someone else understand exactly what you want based on this task? Yes = Good.

✓ Audience is defined

If AI creates this, does it know exactly who it's for? Yes = Good.

✓ Tone is unmistakable

Could someone read your tone requirements and match your voice? Yes = Good.

✓ Format is specific

Does your prompt describe exactly how the output should be organized? Yes = Good.

✓ Requirements are listed

Is there a "must include" section covering everything non-negotiable? Yes = Good.

✓ Nothing is vague

Is there anything AI could interpret multiple ways? If YES—rewrite it clearer.

✓ You can test it right now

Could you paste this into ChatGPT and expect good results? Yes = DONE.

🚀 Your Challenge Today

Here's exactly what to do:

Pick ONE task you do regularly. Blog post? Email? Product description?
Answer the Five Questions: WHAT? WHO? HOW (sound)? WHAT (include)? HOW (organized)?
Build your prompt layer by layer. Basics → Audience → Voice → Format → Requirements.
Use the bracket thinking pattern. For each section, ask "what does AI need to know?"
Check the 7-point checklist. Is everything clear and specific?
Test it. Paste it into ChatGPT right now.
Reply: What task did you pick? Did your prompt work? What surprised you?

Important: First try might not be perfect. That's normal. Test. See what comes back. Adjust. Test again. That's how you learn.

⚡ Why This Matters

Days 1-2: You used templates.

Today: You learned to build templates from scratch.

Tomorrow: You'll refine and iterate them.

By day 7: You'll have a complete workflow that saves you hours every week.

👀 What's Coming Tomorrow

Day 4: Iterating and refining. You'll learn how to take feedback from AI's output and improve your prompt. How to layer techniques. How to handle edge cases.

But first—go build your own prompt from scratch. Reply and show me what you created.