How to Prompt Better Than 99% of ChatGPT Users

Most people use ChatGPT like a basic search bar or a simple Q&A tool, getting generic and mediocre results. By treating ChatGPT as a strategic partner and establishing a clear structure, you can unlock far better outputs.

Here is a step-by-step framework to master AI prompting and outperform almost every standard ChatGPT user.

---

The Prompting Framework

[01. Use Projects] ➔ [02. Thinking Mode] ➔ [03. Web Search]
         │
         ▼
[06. State Boundaries] ◄── [05. Success Brief] ◄── [04. Real Example]
         │
         ▼
[07. Use Critique] ➔ [08. Correct Fast] ➔ [09. Reset Chat]

---

Step 01: Use Projects, Not Chats

Stop starting isolated, context-free chats for every quick thought.

• Create dedicated 'Projects': Set up specific projects for distinct tasks or domains.
• Upload core files: Include relevant reference materials, background context, and data.
• Write system instructions: Define the persona, rules, and expectations once.

Key Rule: Set it up once. Done.

---

Step 02: Turn On Thinking Mode

Before writing your prompt, turn on Extended Thinking (or Reasoning mode).

• Forces ChatGPT to reason through complex logic and structure its thoughts before delivering an answer.
• Reduces surface-level responses and halluncinations on nuanced tasks.

Key Rule: Thinking on. Always.

---

Step 03: Turn On Web Search

Never rely purely on static training data for facts or recent events.

• Turn on Web Search to force ChatGPT to ground its answers in real-world sources.
• Prevents the model from making up facts or referencing outdated information.

Key Rule: Search on. No lies.

---

Step 04: Upload a Real Example

Stop describing what you want in paragraph form—show the model instead.

• Find reference content: Locate an existing example that matches your target format or style.
• Format as Markdown: Download or convert the example into clear markdown text.
• Upload as reference: Give ChatGPT a concrete anchor for tone, structure, and quality.

Key Rule: Show, don't describe.

---

Step 05: Fill Out Your Success Brief

Define what "good" looks like before you ask for an output by answering four quick questions:

1. Type of output + length: What is it? How long should it be?
2. Recipient's reaction: What should the target audience think, feel, or do?
3. Does NOT sound like: Generic AI? Too casual? Jargon-heavy?
4. Success means: Do they sign? Approve? Reply? Act?

Key Rule: Know what "good" looks like before you ask.

---

Step 06: State Goal and Boundaries

Tell the AI what you need using a structured boundary prompt:

"I need [output type] that [does what]. This is for [audience], where [what matters]. Focus on [priority], avoid [anti-priority]. After reading, they should [action]."

Key Rule: Don't tell AI how to do it. Tell it what you need.

---

Step 07: Use ChatGPT to Critique

Instead of asking ChatGPT to write the final piece immediately, use it as a harsh editor first.

Prompt ChatGPT to Critique:

"Your job is NOT to rewrite. Your job is to CRITIQUE. Read this and identify: 1) The 3 weakest lines and why. 2) Where the reader loses interest. 3) What's missing. Grade each section A/B/C/D/F."

Key Rule: You fix it. Then repeat.

---

Step 08: Correct Fast. Be Direct.

When the AI makes a mistake or strays off course, fix it immediately without overly polite filler.

• State the correction directly: "That's wrong, X is actually Y. Fix it and continue."
• Quote the exact line or phrase you are referring to.

Key Rule: Short. Direct. Move on.

---

Step 09: Reset When It Gets Messy

After 8–10 back-and-forths, chat quality naturally drops due to context clutter.

1. Copy the best output produced so far.
2. Start a fresh new chat session.
3. Paste the clean output into the new chat.
4. Prompt: "Here's where we are. Continue from here."

---

Summary Checklist

Step	Action	Core Mindset
01	Use Projects	Single setup for persistent context
02	Thinking Mode	Force reasoning before response
03	Web Search	Demand factual, grounded sources
04	Real Example	Show concrete references
05	Success Brief	Define criteria for success
06	Goal & Boundaries	State requirements clearly
07	Use Critique	Act as editor before writer
08	Correct Fast	Direct feedback without fluff
09	Reset Chat	Keep context clean and effective