The 6-Part Prompt Formula Behind My Best AI Outputs

The core structure I use to turn any idea into a great Prompt.

Most people write prompts like they’re texting a chatbot.

But to unlock LLMs’ full potential?

You need to think like a prompt architect.

That means giving clear roles, outcomes, tone, and structure just like briefing a real teammate.

This guide will show you exactly how to do that with real-world examples and side-by-side comparisons of bad vs great prompts.

A bad prompt wastes your time.
A smart prompt saves it.

Let’s dive in.

What Makes a Good Prompt?

A high-quality prompt is like a smart brief. It gives GPT just enough structure to hit the mark.

Use this simple structure:

• Context → What’s the scenario or background?

• Role → Who should GPT act as?
  (e.g., CEO, Copywriter)

• Task → What do you want it to do? (e.g., write a thread, summarize a blog)

• Format → How should the response look? (e.g., table, list, paragraphs)

• Clarify → Any boundaries? (word count, tone, audience)

• Refine → Add examples to improve output

This formula helps GPT give human-quality results with way less effort.

---

What You Need

• Any LLM (i.e ChatGPT , Claude or Gemini)

• Relaxed mind

• One topic you want to explore

---

▸ Step 1:

Start With a Clear Task

Why: Vague inputs lead to vague outputs.

Bad Prompt:

“Write about productivity.”

Good Prompt:

“You are a productivity coach. Write a 200-word LinkedIn post for startup founders on how to structure their mornings. Format it as 5 punchy bullet points.”

---

▸ Step 2:

Assign a Role

Why: Role shifts tone and vocabulary.

Bad Prompt:

“Explain AI to beginners.”

Good Prompt:

“Act as a friendly AI teacher. Explain how ChatGPT works to a group of high school students using simple analogies.”

---

▸ Step 3:

Define the Format

Why: Prevents walls of text or scattered ideas.

Bad Prompt:

“Summarize this article.”

Good Prompt:

“Summarize the article in 5 bullet points for busy managers. Keep each under 15 words.”

---

▸ Step 4:

Add Context Early

Why: Memory helps, but context guides output instantly.

Prompt:

“We're writing onboarding emails for new Canva users. Our tone is friendly, confident, and encouraging. Now write a welcome email as if you’re a helpful coach.”

---

▸ Step 5:

Give Examples (Optional but Powerful)

Why: Shows GPT what “good” looks like.

Good Example → “Welcome aboard! Here’s how to get started fast.” 
Bad Example → “Thank you for registering for our software.” 
Now write 3 new variants that match the good tone.

---

▸ Step 6:

Refine & Save What Works

Why: One great prompt can be reused forever.

• Save your best prompts in Notion, Google Docs, or anywhere

• Version them for different tones (casual, executive, Gen Z)

• Create a “prompt vault” for common use cases

---

Done!

Now You’re Prompting Like a Pro.

Keep this formula in your back pocket:
Context + Role + Task + Format + Clarify + Refine
Use it consistently, and your prompts will work harder than most humans.

Bonus tip: Keep a “best prompts” doc as you go.
You’ll build a personal toolkit you can plug into any AI task.

See you in a few days,
Aniket Chhetri

---

🤔💭 What’s the biggest mistake you’ve made while prompting AI?

Let’s talk in the comments section below.