How to Write Better ChatGPT Prompts (Stop Getting Generic AI Answers)

The Value of the Tool Lies in the User of the Tool

Remember this: the value of a tool lies in the user of that tool. Prompt creation is probably one of the most important skills you can learn when it comes to experiencing the potential of AI. A good prompt produces high quality results you can actually use. A bad prompt leads to generic results.

All of these tools we have today, ChatGPT, Gemini, Llama, you name it, are generative AI. They depend on your input to give you an output. So if you're an intelligent prompt engineer, you get very intelligent responses. If you're not very smart with it, you're not going to get a lot of smart responses back. That's the whole game.

Learning this skill is incredibly important, so let me give you a few examples of how you can take this tool to the next level for yourself.

The Bakery Test: Why 'Pretty Good' Prompts Are Actually Bad

Here's a bad prompt: 'Give me 20 ideas for social media posts I can make for my bakery business.' Say you're trying to build a following for your local bakery shop. It's going to give you 20 responses and you're going to feel like, wow, that's amazing. Behind the scenes videos, customer creations, meet the team. That's pretty good, right? So why am I calling this bad?

Because you have to go deeper. If you ask ChatGPT a stupid request, you're going to get stupid responses. If you give it intelligent requests, you're going to get intelligent responses.

Now take that small, simple prompt and go somewhere much deeper: 'I own a bakery in Clearwater, Florida. I sell custom cakes, pastries, and cookies. Create 20 social media post ideas that include details on the type of post (image, video, or text), specific instructions for what the creative should look like, and captions and hashtags for each. The primary purpose of these posts is to encourage more people to come in and buy my products.' When I ask something like that, I get way more detailed responses. Things I can take, copy, paste, and make more things with. Things that create real impact.

Element One: The Objective

Clearly state what you want ChatGPT, or any AI tool you care to choose, to do or produce. Set the objective from the start. This is the most important part of a prompt. Without a clear objective, it will not know how to respond.

And these tools can do a lot: generate ideas, translate hundreds of languages at your fingertips, provide insights, edit content, conduct research, solve problems, draft emails, explain concepts, identify trends, recommend tools, define words, fix errors, and a whole lot more. Effective prompts use words that clearly tell it what you want.

Compare the two. Don't type 'food for weight loss.' Instead: 'Create a 30-day meal plan that will help me lose weight by incorporating foods that are known to help weight loss.' The first gets you a list that feels fine. The second gets you day one breakfast, lunch, dinner, even a snack, then day two, and so on. Sophisticated responses you can actually apply and implement. More done in less time.

Element Two: Context

It's not just what the objective is. It's what context you provide to help accomplish that objective. Context is the background details that ensure the tool understands your request accurately. The more you feed it, the more quality you get back.

What counts as context? Who the intended result is for (business owners, for example). Location (New York City). Event details (a wedding ceremony). Purpose (to inform about recycling). Audience knowledge (experts in the field). Type of document (a formal business letter). You get the idea.

Look at the difference. 'Create a course outline about artificial intelligence' versus 'Create a course outline about AI and how it applies to business owners and their employees.' Now I get it lesson by lesson, module by module, built exactly for helping business owners and their employees get more production and value from AI.

Same with email. 'Write an email to promote my upcoming webinar about diabetes' is very incomplete, very short. Instead: 'Write an email to promote my upcoming webinar about diabetes. The webinar will be on August 15th at 3:00 PM Pacific time. I will be covering new information I've recently discovered about the subject,' and so on. The more data you give it, the better the output it gives you.

Element Three: Examples

Include an example of the desired output. If you have something you've already written that you like, the way you describe things, the words you use, how you avoid complicated words, use it as an example and the AI will adjust to that particular way of communicating.

Not every prompt needs an example, but they help you produce a more specific result. You can specify a writing style: write like a best-selling author you admire, like Hemingway. Text formatting: bullet point format. Content structure: follow a problem-solution format.

Here's how one sentence changes everything. 'Create copy for a landing page to promote my upcoming webinar about TikTok ads on August 15 at 5:00 PM Eastern time.' Now add: 'Format the copy with this structure in mind: Problem, Agitate, Solve.' That extra sentence produces something entirely different. Or take 'Write a script for kids about the importance of brushing your teeth' and add 'do it in the style of Stan Lee,' the legendary creator of Spider-Man. You get twists and angles you wouldn't have otherwise.

Reverse Prompt Engineering: Make the AI Coach You

Quick recap: an effective prompt has an objective, context, and examples. Put those together and you have a prompt that gets real value from the tool. Now here's the move that ties it all together: get ChatGPT to help you structure the prompt itself.

This is the prompt I use: 'You are a professional at creating effective ChatGPT prompts. Ask me questions one by one, not all at once, to get the information you need. For each question, give me a list of general ideas and examples so I can better answer you. Ultimately, the prompt should include a clear objective, context, and examples if applicable.'

I call this reverse prompt engineering, because you're using ChatGPT to teach you how to ask it questions so it can give you what you want. When I ran it, it offered me categories: marketing strategy creation, like a YouTube content plan to grow my audience by 100,000 followers in six months. Creative content generation, like a persuasive product description for a new supplement line. Business systems and SOPs, like a standard operating procedure for onboarding new employees at my agency. Data and financial insight, like analyzing cash flow and suggesting improvements.

So I took one and said, let's see what you can do for me. I want to grow my YouTube channel by 100,000 followers in six months. What's the game plan? And immediately it started asking me questions: what's the core theme or niche of your channel? Now I have a coach working with me, asking questions so it understands what I need, so it can guide me. That's prompt engineering at its best.