Become a ChatGPT Power User

Master 7 prompting techniques that turn ChatGPT into a precision tool.

Introduction

30 Second Summary

You have probably had the experience of asking ChatGPT a question and getting back an answer that is technically correct but completely generic. It reads like it could have been written for anyone, because it was.

In this project, you will tackle seven rapid-fire challenges that each teach one named power-user technique for driving ChatGPT conversations. You will combine these techniques with everything you learned in Projects #1-3 to transform a lazy one-sentence prompt into a personalized, battle-tested result.

What You'll Build

You will take a real problem you are actually dealing with, run seven named techniques on it inside a single ChatGPT conversation, and watch the answer transform from generic filler into a personalized action plan.

By the end of this project, you'll have:

The power to control any conversation by making ChatGPT interview you first, generate competing approaches, and challenge its own ideas before you commit to a solution.
The ability to turn ChatGPT into its own editor so it grades and rewrites its own drafts. Show it one example and watch it lock onto the pattern across new content.
A before-and-after comparison on a real personal problem where you chain all seven techniques together and see a one-sentence prompt produce a result that is specific to your life, not a generic template.
Secret Mission: Build a personal ChatGPT Power User Cheat Sheet capturing your favorite techniques, reusable prompts, and personal rules you will actually open every time you start a ChatGPT conversation.

Are there any prerequisites?

This is Project #4 in the ChatGPT Power User Curriculum. Complete Projects #1-3 (personal context setup, structured prompting, and model selection) before starting this one.

Before We Start

Before you dive into the challenges, take a moment to lock in what you are about to build and why it matters.

Take Your Before Shot

Before you learn any new techniques, you need proof of where you start. A baseline captures your current results so you can measure real improvement instead of just feeling like things got better.

In this step, you will send a deliberately lazy prompt to ChatGPT. The response becomes your Before Shot, a snapshot of what happens when you give ChatGPT nothing to work with.

In this step, get ready to:

Open ChatGPT and start a fresh conversation.
Send a lazy prompt and observe the generic response.
Bookmark the response as your Before Shot baseline.

Open ChatGPT and start a new conversation

Open chatgpt.com in your browser.
Start a new conversation.

Which plan am I on?

Plus users will see a reasoning slider in the model picker. On web, the model picker appears inside the message composer. On mobile, it sits at the top of the conversation. Options include Instant, Medium, High, and Extra High.

Free users have unlimited text chats with GPT-5.6 Luna. Look for the Think button for harder questions.

Go users can click + in the chat entry box to find the Thinking option.

All plans work for every challenge in this project.

If you are on Plus, set your reasoning to Instant for now.
If you are on Free or Go, leave the default setting.

Send the lazy prompt

What do you think ChatGPT will produce when you type a single sentence with no context about who you are, what you fear, or what you specifically need?

Send this exact prompt with zero setup, zero context, and zero technique:

Help me get better at public speaking.

Read the response carefully.

The response is a generic list of public speaking tips that could apply to literally anyone. It reads like a self-help article, not a personalized plan.

Why is the response so generic?

When you give ChatGPT nothing specific to work with, it fills in the blanks with the most common assumptions. This is what most people accept and move on from.

The response does not mention your experience level, your specific fears, the speaking context you face, or what "better" means to you.

Keep this conversation open. You will use it for every challenge so ChatGPT builds context as you go.
Mentally bookmark this response as your Before Shot.

Response looks different than expected?

ChatGPT produces different responses each time, so your exact wording will differ. The key observation is whether the advice is generic enough to apply to anyone.

If the tips feel surprisingly personal despite your one-sentence prompt, you may have custom instructions active from a previous project. The observation still holds: your prompt gave ChatGPT nothing specific to work with.

Why does my ChatGPT response seem personalized even though I used a lazy prompt?

Your Before Shot is locked in. Next up, you will flip the dynamic by making ChatGPT ask you the right questions before it starts solving.

Control the Conversation

Your Before Shot proved something important. When you send ChatGPT a vague request, you get advice that could apply to literally anyone.

These three challenges close that gap. You will take control of what ChatGPT knows, what it considers, and how critically it thinks before you accept a single word.

In this step, get ready to:

Use the Interview Flip to get personalized advice instead of generic tips.
Demand multiple approaches with the Three-Door Test.
Stress-test your chosen approach using the Devil's Advocate.

Try the Interview Flip

Most people describe their problem and hope ChatGPT fills in the blanks correctly. The Interview Flip does the opposite. You make ChatGPT ask you the right questions before it writes a single word of advice.

In your open conversation (the one with your Before Shot), send this prompt:

I want to get better at public speaking. Before you give me any advice, interview me. Ask me 5 specific questions that will help you give me a personalized plan instead of generic tips.

What does the Interview Flip do?

Instead of guessing your situation, ChatGPT now has to ask. The five questions force it to learn your experience level, your specific fears, your upcoming events, and what "better" actually means for you.

This is different from the structured prompting you learned in Project #2. There, you provided context upfront. Here, you let ChatGPT figure out which context matters most.

Read the five questions ChatGPT asks you.
Answer all five questions honestly in a single response. Use real, specific detail from your life.

Use your Project #1 skills

In Project #1, you learned to give ChatGPT personal context. Apply that here. Instead of "I get nervous," try "I get nervous presenting quarterly results to my team of 12 because I worry about follow-up questions." Specificity in your answers drives specificity in ChatGPT's advice.

After you send your answers, ChatGPT generates a personalized public speaking plan based on what you shared.

Before you look back, predict: how different do you think this plan is from your Before Shot at the top of the conversation?

Scroll up to the top of your conversation and re-read your Before Shot.
Scroll back down and read the Interview Flip result.

The Before Shot was a generic list anyone could follow. The Interview Flip result should reference your specific answers and read like advice written for you, not a self-help article.

Personalized plan still feels generic?

Check that your answers included specific, personal detail. Vague answers produce vague advice.
If your answers were specific but the plan is still broad, send a follow-up: "Your advice is still too generic. Use the details I gave you about [mention one of your answers] to make it specific to my situation."
Why is my Interview Flip response still generic even after I answered the questions?

Run the Three-Door Test

Most people accept the first approach ChatGPT offers. Power users demand options. The Three-Door Test forces ChatGPT to generate multiple distinct approaches so you can choose the strongest one yourself.

In the same conversation, send this prompt:

Give me 3 completely different approaches to improving my public speaking. Make each one a different philosophy, not three variations of the same advice. For each, name one strength and one trade-off.

Why demand multiple approaches?

ChatGPT's first answer is optimized for being safe and broadly helpful. Asking for three philosophically different approaches forces it to explore angles you would never have considered on your own.

The "strength and trade-off" requirement prevents ChatGPT from just listing three similar ideas with different labels. Each option has to stand on its own merits.

Read all three approaches.
Pick the one that resonates most with you.
Tell ChatGPT which approach you chose and why.

You should see three approaches that feel genuinely different. One might focus on repetition, another on mindset, another on audience connection. If they all sound like variations of "practice more," that is not three doors.

All three approaches feel too similar?

Send a follow-up: "These three approaches are too similar. Give me three that are philosophically different. One could focus on mindset, one on technique, and one on environment. Make them disagree with each other."
The word "philosophy" in the original prompt is doing heavy lifting. If ChatGPT ignored it, repeating it in the follow-up usually fixes the problem.
Why are my Three-Door Test results too similar despite asking for different philosophies?

Trigger the Devil's Advocate

ChatGPT defaults to agreeing with you. It is designed to be helpful, which means it rarely pushes back on your choices. The Devil's Advocate technique forces it to argue against your pick so you can stress-test your decision before committing.

In the same conversation, send this prompt (replacing the placeholder with the approach you picked in the Three-Door Test):

I've picked approach [[YOUR_APPROACH="the number or name of the approach you chose"]]. Now challenge it. What are the strongest arguments that this approach might not work for someone in my situation? Be specific, not vague.

Why force ChatGPT to push back?

Left to its own preferences, ChatGPT will validate whatever you say. That feels good but does not help you think. Forcing a devil's advocate response exposes blind spots in your thinking that you would not have found on your own.

Read the pushback carefully. ChatGPT should name specific reasons your chosen approach might fail for someone in your situation, not vague disclaimers.

Pushback feels too gentle?

ChatGPT sometimes pulls its punches. If the criticism reads like polite hedging instead of a real challenge, send this follow-up:

Don't be polite about it. What would a skeptic say?

This override usually produces sharper, more useful criticism. The goal is not to feel attacked. The goal is to find the real weaknesses in your plan before you commit to it.

How do I get ChatGPT to give me genuinely critical pushback instead of polite hedging?

Think back across all three challenges. Which single technique changed the quality of ChatGPT's responses the most for you?

Your conversation now holds a visible transformation. The Before Shot at the top is a generic list. The Interview Flip result references your specific answers. The Three-Door Test gave you options you never would have asked for. The Devil's Advocate poked holes in your thinking. That is what controlling the conversation looks like.

You have taken control of the input side of the conversation. Next up, you will make ChatGPT evaluate and improve its own output.

Make ChatGPT Its Own Editor

In the last step, you took control of the conversation. You made ChatGPT interview you, generate distinct options, and challenge your thinking before committing to an approach.

Now for a different problem. Most people accept ChatGPT's first draft and move on. These next two challenges make ChatGPT evaluate its own output and improve it, so you get second-draft quality without diagnosing what went wrong yourself.

In this step, get ready to:

Use the Self-Critique Loop to make ChatGPT draft, critique, and rewrite in a single prompt.
Use the Pattern Lock to make ChatGPT infer and replicate a format from one example.

Challenge 4: The Self-Critique Loop

ChatGPT can grade its own work. The trick is combining draft, critique, and rewrite into a single prompt so ChatGPT spots its own weaknesses and fixes them before you even need to read the first version.

Send this prompt in your conversation to see the Self-Critique Loop in action:

Write a 100-word summary of why someone should learn to cook at home instead of eating out. Then critique your own summary: what is weak, generic, or unconvincing? Finally, rewrite it fixing every issue you identified.

What does this prompt do?

The first instruction ("Write a 100-word summary") produces the raw draft.
The second instruction ("critique your own summary") forces ChatGPT to identify specific weaknesses instead of leaving that work to you.
The third instruction ("rewrite it fixing every issue") ensures the critique becomes an action plan for a better version, not just commentary.

Watch for three distinct sections in the response: the original draft, the critique, and the rewrite. The rewrite should be noticeably stronger than the draft.

Sometimes ChatGPT pulls its punches. If the critique reads like a polite suggestion rather than a real assessment, push harder.

Send this follow-up if the critique feels too soft:

Your critique was too polite. Find 3 real weaknesses and rewrite harder.

The updated critique should name specific problems. Vague observations like "could be better" or "might not resonate" are signs ChatGPT is still being too gentle.

Critique still too vague?

Check that your original prompt included all three instructions: draft, critique, and rewrite. If you only asked for a "summary and feedback," ChatGPT may combine sections instead of separating them.
If the rewrite barely changed, try telling ChatGPT: "The rewrite is almost identical to the draft. Be bolder with your changes."

Why isn't ChatGPT's critique specific enough, and how can I push it to find real weaknesses?

Challenge 5: The Pattern Lock

Explaining exactly what format you want can take more effort than writing it yourself. The Pattern Lock skips that problem entirely. You show ChatGPT one example, and it infers the rules on its own.

Before you send the next prompt, make a prediction. You are about to give ChatGPT a single note-taking example with no instructions about the format rules. Do you think it will match your structure exactly, or default to its own format?

Send this prompt in your conversation to test the Pattern Lock:

Here is an example of how I like to take notes:

Topic: Morning routines
Key idea: Small habits compound. A 1% improvement in your morning routine creates a 37x difference over a year.
So what?: Start with one anchor habit tomorrow. Mine is making the bed before checking my phone.

Now write notes in exactly this format for the topic: Time management.

Why does one example work?

Notice what you did not do. You never described the format rules. You never said "use three sections" or "end with a personal action step." ChatGPT inferred all of that from the pattern in your example.

This technique is called few-shot prompting. Even a single example often outperforms a paragraph of instructions because the model locks onto structure, tone, and length all at once.

Check the result against your example. It should match the same three-section structure: Topic, Key idea, and So what? with a personal action step at the end. You did not have to explain the format. ChatGPT locked onto it from the example alone.

Output doesn't match the format?

Make sure your example was formatted clearly with line breaks between sections. If you pasted it as a single paragraph, ChatGPT may not detect the structure.
If the structure is close but the tone is off, follow up with: "Match the casual, personal tone of my example. Keep the 'So what?' action step specific to me."

ChatGPT isn't matching the format I showed it. How can I get it to replicate my exact structure?

Scroll up in your conversation and compare these two challenges to the Before Shot and the Interview Flip. You now have two more ways to get dramatically better results: making ChatGPT grade its own work, and showing it what you want instead of explaining it.

You have five techniques in your toolkit now. Next, you will learn to reshape what ChatGPT has already built and surgically improve specific parts without starting over.

Shape and Refine Like a Pro

In the last step, you made ChatGPT its own editor. The Self-Critique Loop showed that ChatGPT can grade and rewrite its own work when you push it. The Pattern Lock proved that one example is more powerful than a paragraph of instructions.

Those techniques improved the quality of ChatGPT's output. But what happens when the output is good but in the wrong shape? Power users rarely start over. They reshape what already exists and surgically improve the parts that need work.

In this step, get ready to:

Transform an existing response into a completely different format using The Format Shift.
Surgically improve one weak section of a response without rewriting the rest using The Remix.

Challenge 6: The Format Shift

The Interview Flip back in Step 2 gave you personalized public speaking advice. That advice was solid, but it landed as a block of recommendations. A list of tips is not the same as a plan you can act on day by day.

The Format Shift takes an existing response and transforms it into a completely different format. The substance stays the same. Only the shape changes.

In your existing conversation, send this prompt to reshape your Interview Flip advice into a day-by-day action plan:

Take the public speaking advice you gave me after the interview and transform it into a 5-day action plan. Each day should have one specific task I can complete in under 15 minutes. Include a way to measure whether I did it.

What does The Format Shift do?

ChatGPT already did the hard work of personalizing your advice during the Interview Flip. This prompt asks it to restructure that same substance into a more actionable shape.

You are not asking for new advice. You are asking ChatGPT to repackage what it already gave you into a format you can actually follow.

Read through the 5-day plan. Each day should have a specific task, a time commitment under 15 minutes, and a way to measure success.
Scroll up and compare the plan to your original Interview Flip advice from Step 2. The substance should be the same, just restructured into daily actions.

Plan looks generic or doesn't match your earlier advice?

ChatGPT may have lost track of the Interview Flip in a long conversation. Scroll up, copy a key detail from your Interview Flip result, and add it to a follow-up message like: "My interview advice mentioned [this specific detail]. Make sure the plan reflects that."

If the tasks are too vague ("practice speaking" instead of a concrete action), follow up with: "Make each task something I can do in one sitting with a clear start and end."

Still stuck? Help me get my Format Shift to produce a specific, personalized plan.

Challenge 7: The Remix

You have a 5-day plan, but not every day will be equally strong. Some tasks might feel vague or too easy. Most people would regenerate the entire plan from scratch. Power users fix only the part that needs work.

Read through your 5-day plan from Challenge 6.
Find one day where the task feels weak, vague, or too generic.
Send this prompt in the same conversation, filling in the day number you picked:

Day [[DAY_NUMBER="the day number you want to improve"]] is too vague. Make the task more specific, add an example of what success looks like, and keep everything else exactly the same.

What does The Remix do?

The Remix targets one specific section of a larger response for improvement. By explicitly saying "keep everything else exactly the same," you prevent ChatGPT from rewriting the entire response.

This is how you iterate without starting over. You keep what works and sharpen what does not.

Before you read the result, predict: did ChatGPT change only the day you targeted, or did it rewrite the entire plan?

Read the updated plan. Only the day you targeted should have changed. The other four days should be identical to the Format Shift result.
Check that the improved day now has a more specific task and an example of what success looks like.

ChatGPT rewrote the entire plan?

This is a common behavior. ChatGPT sometimes interprets a revision request as permission to regenerate everything. Send this follow-up to correct it:

I said keep everything else the same. Only change Day [[DAY_NUMBER="the day number you want to improve"]].

This is part of the technique. Learning to push back when ChatGPT drifts from your instructions is a core power-user skill.

Need more help? Help me get ChatGPT to change only one section without rewriting everything.

Compare your before and after

All seven challenges are now complete in your conversation. Before you move on, take a moment to see how far you have come.

Think about this: how different is your final 5-day plan from the generic Before Shot at the top of this conversation?

Scroll to the top of your conversation and re-read your Before Shot response.
Scroll down to your Remix-improved 5-day plan.

The Before Shot gave you generic advice that could apply to anyone. Your final plan references personal details from your interview answers, is structured into daily actions with measurable success criteria, and has one day surgically sharpened by The Remix.

You now have the full toolkit of seven named techniques that cover every stage of a ChatGPT conversation. Next up, you will chain them all together on a real problem that actually matters to you.

The Final Boss

You have seven named techniques in your toolkit. Each one gave you a different way to steer ChatGPT conversations toward better results.

Individual techniques are powerful on their own. Real power comes from chaining them together on a single messy problem where the answer actually matters to you.

In this step, get ready to:

Pick a real problem and take your Final Boss Before Shot.
Run the full power-user playbook combining all six techniques in one conversation.
Compare your Before Shot to the final result and name which technique had the biggest impact.

Pick your problem and take the Final Boss Before Shot

This time, the problem is real. Pick something you are actually dealing with right now: a decision you need to make, a project you are stuck on, a skill you want to develop, or a message you need to write.

The stakes matter because they make the before-and-after comparison personal. Something where getting a great answer actually helps you.

Start a new conversation in ChatGPT.
Send the laziest possible version of your request in one sentence. No context, no structure, no technique. Just the raw ask.
Read the response carefully.

This is your Final Boss Before Shot. Notice how the advice could apply to almost anyone. It technically answers the question, but it does not know your situation.

Why a new conversation?

The public speaking conversation from Steps 2-4 has a lot of built-up context. Starting fresh gives you a clean comparison: lazy prompt versus the full playbook, with no prior context helping ChatGPT along.

Run the full power-user playbook

Now chain every technique from this project on the same problem, in the same conversation. Each technique builds on the last.

Start with The Interview Flip. Send this prompt:

Before you help me with this, interview me. Ask the 5 questions that will most change your approach.

Answer each question with real detail from your life. Apply the personal context skills you built in Project #1.
Let ChatGPT generate its personalized response based on your answers.

Next, run The Three-Door Test. Send this prompt:

Give me 3 fundamentally different approaches. For each, name a strength and a trade-off.

Read all three approaches.
Pick the one that fits your situation best and tell ChatGPT which one you chose and why.

Now apply The Devil's Advocate. Send this prompt:

Challenge my choice. What is the strongest reason this approach could fail for me specifically?

Read the pushback. If it feels too gentle, tell ChatGPT to push harder.

Where are you now?

You have gathered context (Interview Flip), explored options (Three-Door Test), and stress-tested your choice (Devil's Advocate). The next three techniques polish the output into something you can actually use.

Apply The Self-Critique Loop. Send this prompt:

Now write your best recommendation based on everything you know. Then critique it and rewrite it stronger.

Read all three sections: the draft, the critique, and the rewrite. The rewrite should be noticeably sharper than the draft.

Apply The Format Shift. Decide which format would be most useful for your problem: an action plan, an email draft, a decision matrix, a checklist, or something else entirely.
Send a prompt asking ChatGPT to transform its recommendation into that format. Be specific about what you want each section to include.

Finally, apply The Remix. Read through the transformed result and find the weakest section.
Send a prompt telling ChatGPT exactly which part is weak, what to fix, and to keep everything else the same.

If ChatGPT rewrites everything

Sometimes ChatGPT ignores the "keep everything else the same" instruction and rewrites the entire output. If that happens, tell it directly: "I said keep everything else the same. Only change the section I named."

Compare your Before Shot to the final result

Before you scroll up, make a prediction. Do you think the final result will be slightly better than your one-sentence Before Shot, or dramatically different?

Scroll to the top of your Final Boss conversation.
Read your Before Shot response first. Then scroll down to your final result and read it.

The gap between the two should be striking. Your Before Shot is generic advice that could apply to anyone. Your final result references your specific situation, accounts for trade-offs, has been stress-tested against its own weaknesses, and lives in a format you chose.

Answer these two questions for yourself:

Why these two questions matter

Naming your strongest technique helps you reach for it first next time. Naming what you would change turns this from a one-time exercise into a repeatable process you will improve every time you use it.

Not seeing a dramatic difference?

If your final result does not feel dramatically better than the Before Shot, check two things. First, did you give real, specific answers during the Interview Flip? Vague answers produce vague advice. Second, did you push back during the Devil's Advocate and Self-Critique Loop, or did you accept the first response?

Try running the playbook again on the same problem, but answer the interview questions with more detail and push ChatGPT harder on its critique.

Why doesn't my final result feel better than my Before Shot?

Secret mission

Build Your Power User Cheat Sheet

Techniques you don't write down become techniques you forget. Turn everything you learned across Projects #1-4 into a personal ChatGPT Power User Cheat Sheet you will actually open every time you start a conversation.

Clean Up Your Resources

This project runs entirely in ChatGPT conversations, so there are no ongoing costs. If you are on a ChatGPT Plus subscription ($20/month), that billing continues regardless of this project.

Resources you used:

ChatGPT conversation (public speaking challenges from Steps 2-4).
ChatGPT conversation (Final Boss challenge with your real personal problem).
ChatGPT conversation (Power User Cheat Sheet scaffold and iterations).

Keep everything running

No action needed. Choose this if you want to revisit your challenge responses, copy prompts, or reference your techniques later.

Your three project conversations stay in your ChatGPT sidebar history at no cost.
Your personal Power User Cheat Sheet stays wherever you saved it (phone note, bookmark, pinned document, or ChatGPT Project).

Pause - I'll come back to this later

Since this project used only ChatGPT conversations, there is nothing to pause or shut down. Your conversations and cheat sheet remain available at no cost whenever you return.

Delete - I don't want to use this again

Remove all project conversations from your ChatGPT history. Your cheat sheet lives outside ChatGPT, so keep or delete it separately wherever you saved it.

Open ChatGPT at chatgpt.com.
In the sidebar, find the public speaking challenges conversation.
Open the conversation menu and delete it.
Repeat for the Final Boss conversation and the Power User Cheat Sheet conversation.

Save your prompts first

Before deleting, consider copying your favorite prompts from the conversations into your cheat sheet. Once a conversation is deleted, its history is gone.

That's a Wrap!

Nice work! You just completed seven power-user challenges and proved what happens when you stop accepting ChatGPT's first draft.

You've learned how to:

Deployed seven named techniques (The Interview Flip, The Three-Door Test, The Devil's Advocate, The Self-Critique Loop, The Pattern Lock, The Format Shift, The Remix) that you can call on by name in any ChatGPT conversation.
Built a before-and-after comparison on a real personal problem that proves these techniques produce dramatically better results than one-shot prompting.
Completed The Final Boss by combining personal context, structured prompting, model awareness, and all seven techniques into a single workflow on a problem that actually matters to you.
Secret Mission: Created a personal ChatGPT Power User Cheat Sheet with your favorite techniques, reusable prompt templates, and personal rules captured for every future conversation.

This completes Chapter 1: Control ChatGPT. You started as someone who typed questions into a box and hoped for good answers.

You now personalize ChatGPT to your context, write structured prompts that get specific results, choose the right model for every task, and drive conversations with techniques most people never discover. Chapter 2: Create With ChatGPT starts with Project #5.

Ready to quiz yourself?