5 Gemini Prompts That Actually Save You Hours Each Week

I'm going to teach you how to prompt Gemini AI like a pro in just five steps. This process works across Claude, ChatGPT, or any large language model you're using. Whether you're new to AI or frustrated with generic, robotic, or inaccurate results, I'll walk you through a simple system that comes directly from Google itself.

I help people with online business and focus on Google Gemini, but you can apply these techniques to any AI platform. I've distilled the key insights from Google's Prompting Guide 101 into five practical steps that will transform how you work with AI. The goal is to stop treating AI like a search bar and start using it as a collaborative partner that understands your business, your audience, and your goals.

Key Takeaways

• Structure your prompts using four components: persona, task, context, and format to get more accurate and relevant responses
• Engage in back-and-forth dialogue with AI rather than one-shot prompting to refine outputs and achieve better results
• Apply these prompting techniques to practical business tasks like homepage optimization and content creation for improved conversion rates

Understanding Large Language Models

How Generative AI Differs from Search

I want you to stop treating large language models like a search bar. There's a little window where you can put in your prompt, but a lot of people are using this with just three, five, or even 10 words like you would do in a Google search.

This is more intelligent than a Google search. I encourage you to join in with a conversation with your large language models and make it a collaborative conversation.

For example, if I wanted to review my website's homepage, I could put "Hey, Gemini, please review my homepage," give it the URL, and share your thoughts. You can go back and forth on this and say things like "I really like number three. Please, can you give me some examples of how I can actually put this into action?"

This is opening dialogue and using the AI in a collaborative way back and forth. You're going to get some better results that way. It gets to know who you are, who you serve, and what your goals are, and it's going to be like a second brain for you to actually achieve that goal a little bit quicker.

Key Platforms: Gemini, Claude, and ChatGPT

I'm going to teach you how to prompt Gemini AI like a pro in just five steps, and this is transferable across Claude, across ChatGPT, whichever large language model you are using. I focused in on Google Gemini because it is Google.

The fastest growing large language model is actually ChatGPT, but because I focus on YouTube and Google, I have focused in on Gemini. You can get exactly the same results with whatever large language model you are using.

Whether you're using ChatGPT, Claude, or even Gemini, the principles I'm teaching apply across all platforms. I'm using the normal Google Gemini, and if you don't have this, you can Google Gemini and you might have to log into Google. I have a pro account because I pay for Google Workspace, but the free Google Gemini account is just as good.

Common User Challenges

If you've just started out in using a large language model like ChatGPT or Gemini, I'm going to walk you through this simple process. If you have used it and you've got generic results or quite robotic results or even inaccurate results, this video is for you.

A lot of people do what's called one-shot prompting, which means that you'll put one prompt in and then you just sit back and then take whatever is the output. But either two-shot prompting or multiple shot prompting means you're having that dialogue and you're iterating on the output.

If I was to go back and forth, back and forth, back and forth, I would ask the AI to sharpen up on some of its outputs. You can engage it more and open this dialogue to improve the results you're getting.

Building Effective Collaboration with AI

Initiating Conversational Interactions

The first thing I want you to do is stop treating AI like a search bar. Many people use the prompt window with just three, five, or even ten words like they would in a Google search. This is more intelligent than a Google search.

I encourage you to join in with a conversation with your large language model and make it a collaborative conversation. For example, if I wanted to review my website's homepage, I could put "Hey, Gemini, please review my homepage." I'd give it the URL and share my thoughts.

When I did this with my homepage at tempp.com, Gemini analyzed my homepage content, website structure, and value proposition. It provided bullet point feedback including:

• Clear value proposition
• Problem solution focus
• Strong call to action
• Excellent social proof
• Clear design

The output indicated my homepage is a highly effective landing page that functions as a powerful funnel clearly designed to generate leads for my free workshop.

Encouraging Two-Way Dialogue

I can engage AI more and open this dialogue as a two-way conversation. After receiving the initial homepage review, I asked "Can you suggest some ways of improving my homepage so that I can get a higher conversion rate of email opt-ins?"

Gemini provided suggestions including:

• Strengthen the above the fold call to action
• More benefit oriented headline
• Sub headline with bullet lists
• Optimize call to action button text and color
• Enhance social proof near the call to action
• Refine frustration section or problem and agitation
• Simplify the final call to action

That was just two prompts from me. You can go back and forth and say "I really like number three. Please, can you give me some examples of how I can actually put this into action?"

This is an example of opening dialogue and using this as a conversation between you and the AI. It gets to know who you are, who you serve, and what your goals are. It's going to be like a second brain for you to actually achieve that goal a little bit quicker.

Using AI in a collaborative way back and forth will get you better results. Setting up two-shot prompting or multiple shot prompting means you're having that dialogue and you're iterating on the output. If I was to go back and forth repeatedly, I would ask the AI to sharpen up on some of its outputs.

Prompt Structuring Using Google's Four-Part Framework

Defining Clear Roles and Tasks

I structure my prompts using Google's recommended framework that includes four distinct components: persona, task, context, and format. This approach gives the AI more understanding of what I want to achieve.

The persona defines who the AI should act as when responding. For example, I might write "You are a marketing expert in the online business space." This tells the AI what perspective to take.

The task specifies exactly what I want the AI to do. I state this clearly, such as "Critically appraise my homepage and provide me with any suggested improvements that I could make." This removes ambiguity about my expectations.

Establishing Context for Precision

The context section provides background information that shapes the AI's response. I include details about who I help, what tools I use, and what my message needs to reflect.

For my business, I might write: "I help founders and experts build intelligent online businesses without employees. I use tools such as YouTube, Google, and Kajabi. My message needs to reflect this."

This context gives the AI specific parameters to work within. It understands my target audience, my methodology, and my business model. The more context I provide, the more tailored and relevant the output becomes.

Formatting Requests for Clarity

The format component tells the AI how to structure its response. I specify whether I want bullet points, paragraphs, tables, or other formatting styles.

A simple format instruction looks like: "Limit your output to bullet points only." This ensures I receive information in the exact structure I need without having to reformat it myself.

I can also request specific types of formatting based on my needs:

• Bullet points for quick scanning
• Tables for comparative data
• Numbered lists for sequential steps
• Paragraphs for detailed explanations

By including all four components in my prompts, I receive significantly better results than when I treat the AI like a search bar with just a few words.

Iterative Prompting Techniques

Utilizing Multi-Shot Approaches

A lot of people do what's called one-shot prompting. This means you'll put one prompt in and then sit back and take whatever is the output.

Two-shot prompting or multiple-shot prompting means you're having that dialogue and you're iterating on the output. If I was to go back and forth, back and forth, back and forth, I would ask the AI to sharpen up on some of its outputs.

Key benefits of multi-shot prompting:

• More refined and accurate results
• Better alignment with your specific needs
• Progressive improvement of AI understanding

Refining Responses Through Back-and-Forth Dialogue

I encourage you to join in with a conversation with your large language models. Make it a collaborative conversation.

You can go back and forth with this to actually give it some more information. If I wanted to say I wanted to position a free opt-in product here from my homepage messaging, you can go back and forth with it just so you can get a little bit more context and it's going to give you that little bit better clarification in the output.

For example, when I reviewed my homepage, I started with a simple prompt: "Hey, Gemini, please review my homepage." After receiving the initial analysis, I followed up with "Can you suggest some ways of improving my homepage so that I can get a higher conversion rate of email opt-ins?"

I could then continue the dialogue by saying "I really like number three. Please, can you give me some examples of how I can actually put this into action?" This opens up dialogue using this as a conversation between you and the AI. It gets to know who you are and who you serve and what your goals are. It's going to be like a second brain for you to actually achieve that goal a little bit quicker.

Incorporating Personalization and Data Input

Leveraging Uploaded Documents

You can upload documents directly into Gemini to give it specific context about your business. This works the same way across other large language models.

When you upload relevant documents, the AI can analyze your actual content rather than working from general assumptions. For example, if you have program documents, marketing materials, or business frameworks, you can include these in your prompts.

The AI will access and unpack the content you provide. It analyzes the structure, dissects key points, and uses this information to generate more accurate outputs that align with your specific needs.

This approach transforms the interaction from generic responses to tailored recommendations based on your actual materials. You're essentially giving the AI your business context in document form.

Integrating Brand Guidelines and Messages

I structure my prompts to include specific brand messaging so the AI understands who I serve and what my goals are. This creates outputs that match my actual positioning.

In the context section of my prompts, I include details like:

• Who I help (founders and experts)
• What I help them do (build intelligent online businesses without employees)
• Tools I use (YouTube, Google, Kajabi)
• My core message positioning

When I told Gemini that I help people build businesses without employees using specific tools, it adjusted its recommendations accordingly. It suggested I explicitly state the no employees benefit and integrate my core tools into the messaging.

The AI can then sharpen audience naming, adjust testimonials to match my target market, and ensure all suggestions reflect my specific positioning. This means the output aligns with my brand rather than providing generic marketing advice.

You can iterate on the messaging by going back and forth with the AI. If the first output doesn't quite capture your brand voice or positioning, you ask it to refine specific elements until it matches your requirements.

Human Oversight and Final Review

Ensuring Quality and Relevance

When you receive output from your AI, you need to actively review what comes back. I don't just accept whatever the AI gives me in that first response.

I go back and forth with the AI to refine the output. This is what I call multiple shot prompting rather than one-shot prompting. One-shot prompting means you put in one prompt and just sit back and take whatever comes out. Multiple shot prompting means I'm having that dialogue and iterating on the output.

I ask the AI to sharpen up on some of its outputs. I might say I really like a specific suggestion and ask for more examples of how to put it into action. This back and forth conversation helps me get better results that are more relevant to my specific needs.

The AI gets to know who I am, who I serve, and what my goals are through this process. It becomes like a second brain to help me achieve my goals a little bit quicker.

Applying the 80/20 Rule

Not every suggestion from the AI will be useful or relevant to my situation. I need to review the output and decide what works for me.

In my homepage review example, the AI suggested I clarify my Kajabi link in my header bar. I like to use really short one-word links, so this is something I would ignore.

I take what makes sense and leave what doesn't. The AI provides options and recommendations, but I make the final decision on what to implement. I focus on the suggestions that will have the biggest impact on my specific goals and ignore the rest.

Applying Pro-Level Prompting to Online Business Tasks

Website Review and Lead Generation

I started by testing a simple prompt with Gemini to review my homepage. I entered "Hey, Gemini, please review my homepage" along with my URL and asked it to share its thoughts. Gemini analyzed my homepage content, website structure, and value proposition.

The output provided bullet points covering clear value proposition, problem solution focus, strong call to action, excellent social proof, and clear design. Gemini concluded my homepage functions as a powerful funnel clearly designed to generate leads for my free workshop.

I then continued the dialogue by asking how I could improve my homepage to get a higher conversion rate of email opt-ins. My homepage is positioned as a lead magnet for lead generation. When people land on my homepage, they understand what I do, who I help, and where to go if they want help.

Gemini provided several suggestions:

• Strengthen the above the fold call to action
• Use more benefit oriented headline
• Add a sub headline with bullet lists
• Optimize call to action button text and color
• Enhance social proof nearer the call to action
• Refine frustration section or problem and agitation
• Simplify the final call to action

This was just two prompts from me. I could continue by saying I liked number three and asking for specific examples of how to put that into action. This demonstrates opening dialogue and using AI in a collaborative way back and forth.

Optimizing Conversion-Oriented Content

I applied the four main areas from Google's prompting guide 101 to structure my prompts more effectively. These four areas are:

I restructured my homepage review prompt using this framework:

Persona: You are a marketing expert in the online business space.

Task: Critically appraise my homepage and provide me with any suggested improvements that I could make.

Context: I help founders and experts build intelligent online businesses without employees. I use tools such as YouTube, Google, and Kajabi. My message needs to reflect this.

Format: Limit your output to bullet points only.

The output became more detailed and specific. It covered headline alignment, clarity, problem identification, framework, methodology testimonials, tool specificity, call to action, and audience targeting.

When I asked what I should improve, it suggested explicitly stating the no employees benefit and integrating my core tools like YouTube and Google into the how section. It recommended elevating the automation proof with testimonials identifying the automations I help people implement in their online business.

It also suggested sharpening audience naming in my testimonials to identify them as founders or experts since these are the people I help. I could go back and forth to give it more information, like positioning a free opt-in product from my homepage messaging.

This is what's called multiple shot prompting rather than one-shot prompting where you put one prompt in and sit back to take whatever output you get. I would ask the AI to sharpen up on some of its outputs by continuing the dialogue.

Accessing Further Learning and Resources

I strongly encourage you to Google "prompting guide 101, Google Gemini" to find the handbook released by Google Workspace along with Gemini. This is the October 2024 edition titled as a quick start handbook for effective prompts. You might need to sign in or leave your email to get a copy, but this guide is my primary recommendation if you take one thing away from learning about AI prompting.

If you read this guide cover to cover, you'll become a much better prompt engineer. I have distilled the key steps from this handbook, but accessing the full document will give you comprehensive training directly from Google itself.

Key Resources to Access:

• Prompting Guide 101 - The official Google Gemini handbook
• Google Gemini - The free account is just as good as the pro version for basic prompting
• Google Workspace Documentation - Additional context for using Gemini in professional settings

The handbook contains detailed examples throughout the document about the four main areas of effective prompting. I zoomed in on the specific section that outlines persona, task, context, and format as the core structure for writing effective prompts. This framework appears multiple times with different examples you can reference.

You can apply the techniques to whichever large language model you're using, whether that's Claude, ChatGPT, or Gemini. The principles are transferable across all platforms. I focus on Google Gemini because I work with YouTube and Google, but you can get exactly the same results with whatever large language model you prefer.