AI & Productivity · Practical Guide

How to Write Better AI Prompts.

A practical guide to getting better results from AI using four essential elements, real-world examples, and a reusable prompt framework.

Artificial intelligence can help you write emails, generate content, analyze information, plan projects and automate everyday tasks. But there's one skill that can dramatically influence the quality of the results you get: knowing how to ask the right questions.

You've probably experienced this yourself. You ask an AI tool to write something, and it gives you a generic response that doesn't quite capture what you need. You rewrite your request, add more details, and suddenly the output becomes much more useful.

That's the difference between a vague prompt and an effective one.

Google's Gemini for Google Workspace: Prompting Guide 101 , published in October 2024, offers a straightforward framework for communicating more effectively with AI. Although the guide focuses on Gemini, many of its principles can be applied to other generative AI tools, including ChatGPT.

In this article, we'll explore how to write better AI prompts, the four essential elements of an effective prompt, common mistakes to avoid and practical examples you can use in your everyday work.

What Is an AI Prompt?

An AI prompt is an instruction, question or request you give to an artificial intelligence tool to generate a response.

It can be something as simple as asking AI to summarize a document or as detailed as requesting a complete marketing strategy with a specific audience, tone and output structure.

For example, consider these two prompts.

Basic prompt:

"Write a marketing email."

Improved prompt:

"Act as a B2B marketing specialist. Write a professional marketing email introducing our website redesign services to SaaS startup founders. Highlight the benefits of improved website performance, user experience and conversion rates. Keep the email under 150 words and end with a call to action encouraging the recipient to book a discovery call."

Both prompts request a marketing email, but the second gives AI significantly more information about the desired outcome.

It establishes who the AI should act as, what it needs to accomplish, who the intended audience is and how the response should be structured.

This approach is at the heart of effective AI prompting.

The Four Essential Elements of an Effective AI Prompt

Google's prompting guide identifies four main elements to consider when writing a prompt: persona, task, context and format.

You don't need to include all four in every request, but understanding how they work together can help you generate more useful responses.

1. Persona: Tell AI Who It Should Act As

A persona establishes the role or perspective you want the AI to adopt.

Instead of asking AI to provide a general response, assign it a relevant professional role.

For example:

"Act as an experienced SEO specialist."

Or:

"You are a senior web developer specializing in Webflow, Framer and WordPress."

A persona can help AI tailor its language, priorities and recommendations to a particular area of expertise.

For example, if you're developing a website for a healthcare company, you might ask AI to approach the project from the perspective of an experienced UX designer who understands accessibility and healthcare website design.

The persona should be relevant to your task rather than unnecessarily elaborate.

2. Task: Clearly Define What You Want AI to Do

The task is the action you want AI to perform.

Google's guide emphasizes the importance of including a clear verb or command in your prompt.

Useful task instructions include:

  • Write a blog post.
  • Summarize a document.
  • Analyze customer feedback.
  • Create a project timeline.
  • Compare two website platforms.
  • Identify potential improvements.
  • Rewrite content in a professional tone.

Avoid instructions that leave too much room for interpretation.

Instead of saying, "Help me with my website," specify what you actually need.

For example:

"Review the following homepage copy and suggest improvements to the headline, value proposition and calls to action. Focus on making the messaging clearer for small business owners."

A clearly defined task helps AI understand what a successful response should accomplish.

3. Context: Provide the Information AI Needs

Context is often the difference between a generic response and one that's genuinely useful.

It includes the background information, goals, constraints and supporting materials relevant to your request.

Suppose you're asking AI to develop a content marketing strategy.

Without context, it might recommend generic tactics that don't match your business.

With context, you can explain your services, target customers, marketing channels, budget and business goals.

For example:

"I run a freelance web design and development business specializing in Webflow, Framer and WordPress. My target customers are SaaS startups, digital agencies and established small businesses. I want to attract qualified leads through LinkedIn and organic search without relying heavily on paid advertising."

Now AI has enough background information to develop a more relevant strategy.

You can also provide existing documents, brand guidelines, customer research, website copy and other materials.

When working with tools that support document uploads or integrations, use those resources to give AI information specific to your business.

4. Format: Explain How You Want the Response Presented

The format determines how AI should organize its output.

You might want a short paragraph, detailed article, comparison table, checklist, email, project plan or structured report.

For example:

"Present the results in a table with columns for platform, advantages, disadvantages, estimated cost and recommended use cases."

Or:

"Write the response as a 1,500-word blog post with a compelling introduction, descriptive H2 headings, practical examples, a conclusion and five frequently asked questions."

Defining the format can reduce the time you spend reorganizing or rewriting AI-generated content.

It also makes the output easier to integrate into your workflow.

How to Combine All Four Elements

The real value of this framework becomes clear when you combine the elements into one well-structured prompt.

Imagine you're a small business owner planning to redesign your website.

Instead of asking, "How can I improve my website?", try:

"Act as a conversion-focused UX designer. Review the following homepage content for my consulting business. My target audience is small business owners looking for operational support. Identify weaknesses in the messaging, information hierarchy and calls to action. Recommend five improvements and present them in a table showing the problem, proposed solution and expected benefit."

This prompt gives AI a professional perspective, a defined task, relevant business context and a clear output format.

You can adapt the same structure to almost any professional or creative task.

Six Practical Tips for Writing Better AI Prompts

Beyond the four-part framework, Google's guide recommends several techniques for improving your results.

1. Use Natural Language

You don't need complicated technical language to communicate with AI. Write as though you're explaining a task to a colleague.

Use complete sentences, clearly explain your request and avoid unnecessary jargon.

2. Be Specific Without Making Your Prompt Unnecessarily Long

More words don't automatically produce better results. The goal is to provide the information AI actually needs.

Specify your audience, objective, constraints and any important details, but avoid filling your prompt with unrelated instructions.

3. Break Complex Tasks into Smaller Steps

Instead of requesting an entire business strategy in one enormous prompt, divide the work into manageable stages.

For example, start by asking AI to identify your ideal customers. Next, request a marketing strategy for those customers. Then ask for a content calendar, followed by individual blog posts and social media content.

This approach allows you to review and refine each stage before moving to the next.

4. Improve Your Results Through Follow-Up Prompts

Your first response doesn't have to be your final response.

If AI generates a useful blog post but the introduction feels weak, ask it to rewrite just the introduction.

If a proposed strategy is too expensive, ask for an alternative that fits your budget.

Treat prompting as an ongoing conversation rather than a single command.

5. Provide Examples and Supporting Documents

When possible, give AI examples of the type of output you want.

For instance, if you're writing website copy, provide your existing brand guidelines, customer profiles and examples of messaging you like.

If you're using Gemini with supported Google Workspace features, you can reference relevant files from Google Drive to give the tool additional context.

Avoid uploading confidential client information unless you're authorized to use it with the AI service and have checked the applicable privacy settings.

6. Ask AI to Improve Your Prompt

You can use AI to identify weaknesses in your instructions before asking it to complete a task.

Try this:

"Review my prompt and rewrite it to make the task, context and desired output clearer. Identify any missing information that would help you produce a better result."

You can also ask AI what additional information it needs before starting a complex project.

This is particularly useful for business planning, research and creative projects where your initial request may leave important details unspecified.

Five Practical AI Prompts You Can Use Today

Here are five original examples inspired by the use cases in Google's guide.

1. For content marketing

"Act as a content strategist for a B2B technology company. Create a four-week content calendar targeting startup founders interested in improving their online presence. Include blog topics, LinkedIn post ideas, target keywords and suggested calls to action. Present the calendar in a table."

2. For web design and development

"Act as a senior UX designer and web developer. Review the following website brief and recommend an effective homepage structure. Focus on clear messaging, accessibility, responsive design and conversion-focused calls to action. Explain the purpose of each recommended section."

3. For client communication

"Act as an experienced freelance project manager. Draft a professional email updating my client on their website redesign. The homepage and service pages are complete, but the contact form requires additional testing. Explain the current progress, remaining tasks and next steps. Keep the tone friendly and reassuring."

4. For customer research

"Act as a customer research specialist. Analyze the attached customer feedback and identify the five most frequently mentioned problems. Group similar complaints, distinguish recurring themes from isolated issues and suggest practical improvements. Present your findings in a table."

5. For business planning

"Act as a business consultant specializing in small digital agencies. Develop a 90-day client acquisition plan for a freelance web developer targeting SaaS startups and marketing agencies. Prioritize LinkedIn outreach, referrals and organic content marketing. Include weekly activities, measurable goals and realistic performance indicators."

Common AI Prompting Mistakes to Avoid

Even a well-written prompt can produce disappointing results. These common mistakes are worth watching for.

Being too vague: Requests such as "Create a website" or "Write some content" provide little direction. Explain the specific deliverable and its intended purpose.

Combining too many unrelated tasks: Asking AI to develop a business plan, write an entire website and create a marketing campaign in one request can make the response difficult to evaluate. Separate substantial tasks into stages.

Ignoring your target audience: A technical explanation intended for experienced developers should sound different from an explanation written for someone building their first website.

Accepting the first response without reviewing it: AI can generate inaccurate information, unsupported claims and content that doesn't match your brand. Verify important facts and review the output before publishing or sending it.

Failing to refine your instructions: When the output isn't right, explain what needs to change rather than repeatedly submitting the same prompt.

A Reusable AI Prompt Template

Use this template whenever you need a structured starting point.

Act as a [professional role or area of expertise].

Your task is to [specific action or deliverable].

Background: [relevant business information, target audience and project context].

The objective is to [desired outcome].

Requirements: [important details, constraints, tone and examples].

Present your response as [preferred format].

If essential information is missing, ask clarifying questions before proceeding.

Customize the sections to fit your task. For straightforward requests, you may only need a clear task and a little context.

Final Thoughts: Better Prompts Start with Clearer Thinking

Learning how to write effective AI prompts isn't about memorizing complicated commands. It's about communicating your objectives clearly.

Start with a defined task, provide relevant context and specify the type of output you need. For more complex requests, assign an appropriate role, break the work into smaller stages and refine the results through follow-up questions.

Whether you're using AI to write marketing content, manage projects, develop websites or run a small business, these habits can help you get more useful results while spending less time rewriting generic responses.

The most important step is to remain involved in the process. AI can accelerate your work, but your expertise, judgment and understanding of your audience are what make the final result valuable.

Source: Google, Gemini for Google Workspace: Prompting Guide 101 (October 2024). The framework and prompting principles are adapted from the guide; the web development and business examples are original applications;