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.
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;