Writing Better AI Prompts
Learn how to write clear, specific and effective prompts that produce stronger results from ChatGPT and other AI tools.
Before You Start
This is Lesson 10 of the MoneyOnliners AI Academy. In Lesson 9, you learned how ChatGPT works and how beginners can use it responsibly.
Now you will learn how to write better prompts. Clear instructions can improve relevance and structure, but factual accuracy still needs independent checking.
Quick Answer
A useful AI prompt states the task, relevant context, requirements and output format. A role is optional when a particular perspective or tone helps. Good prompts are specific, realistic and easy to evaluate. You do not need complicated language. You need clear instructions, relevant background and a defined result.
Learning Objectives
- Understand what an AI prompt is.
- Use a simple five-part prompt formula.
- Improve vague prompts through context and constraints.
- Use examples and follow-up questions effectively.
- Review AI output instead of accepting the first answer.
- Prepare for Lesson 11: AI for Writing and Content Creation.
What Is an AI Prompt?
An AI prompt is the instruction, question or information you give to an AI system. The prompt tells the model what you want and provides the context needed to produce a useful response.
A prompt can be one sentence, a detailed brief, a document, an image or a series of follow-up instructions. The best prompt length depends on the task.
For a simple request, a short prompt may be enough. For professional work, you may need to include the audience, goal, tone, facts, restrictions and final format.
“Better prompts do not sound more complicated. They make the task easier to understand.”
— MoneyOnliners AI AcademyPrompt writing is not about finding magic words. It is about communicating clearly, just as you would when briefing a human assistant.
The Five-Part Prompt Formula
Use these five parts as a flexible checklist, not a mandatory formula. Simple tasks may need only a clear instruction. Add context, examples and constraints when they solve a specific problem; a longer prompt is not automatically better.
Role
Optionally specify a perspective or tone. A role does not confer professional qualifications.
Task
State exactly what you want completed.
Context
Explain the audience, situation and purpose.
Requirements
Include limits, tone, facts and quality standards.
Format
Specify how the final answer should be organized.
Complete Example
“Act as a beginner-friendly business editor. Rewrite this product description for first-time buyers. Use simple language, avoid exaggerated claims, keep it under 120 words and return one paragraph followed by three benefit bullets.”
Part 1: Give the AI a Useful Role
A role can guide the perspective or tone. Use it when that adds useful direction, and omit it when the task is already clear. “Act as an expert” does not make a response accurate or turn the model into a licensed professional.
- Act as a beginner-friendly teacher.
- Act as a professional editor.
- Act as a customer-support specialist.
- Act as an SEO content strategist.
- Act as a project manager.
Do not use a role to create false authority. For high-stakes topics, AI should not be treated as a licensed professional.
Part 2: Define the Task Clearly
Vague tasks create vague results. Tell the AI exactly what action to perform.
| Weak Task | Improved Task |
|---|---|
| Help me with this. | Summarize this report into five key findings and three action items. |
| Write something about freelancing. | Create a 200-word introduction for beginners explaining what freelancing is. |
| Fix my email. | Rewrite this email to sound professional, polite and direct. |
| Give ideas. | Generate ten blog-post ideas for beginner freelance writers. |
Part 3: Add Relevant Context
Context explains why the task matters and who the output is for.
Useful context may include:
- The target audience.
- The reader’s level of knowledge.
- The business or project goal.
- Important background facts.
- The platform where the content will appear.
- Previous work the output should match.
Without context, the AI may choose the wrong tone, level of detail or assumptions.
Part 4: Set Requirements and Constraints
Requirements tell the AI what quality standards and limits to follow.
Length
Specify a word count, sentence count or time limit.
Tone
Choose professional, friendly, persuasive or educational language.
Audience Level
State whether the reader is a beginner, expert or customer.
Must Include
List essential facts, examples or sections.
Must Avoid
Exclude jargon, unsupported claims or certain topics.
Source Rules
Ask for sources where available, then open them and check each important claim yourself.
Part 5: Specify the Output Format
Formatting instructions make responses easier to use.
- Return the answer as a table.
- Use numbered steps.
- Create a checklist.
- Write a headline followed by three paragraphs.
- Provide JSON, CSV or HTML.
- Separate recommendations into beginner, intermediate and advanced levels.
When the format matters, describe it explicitly, then check the result. A request for JSON, CSV or HTML does not guarantee valid syntax; validate machine-readable output before using it in another system.
Good Prompts vs Weak Prompts
| Weak Prompt | Stronger Prompt |
|---|---|
| Write an article. | Write a 700-word beginner article explaining how freelancers can create a portfolio without clients. Use simple language, practical examples and five steps. |
| Give me marketing ideas. | Generate ten low-cost marketing ideas for a local bakery targeting office workers within five kilometers. |
| Summarize this. | Summarize this report into five findings, three risks and three recommended actions for senior management. |
| Create a plan. | Create a four-week study plan for a beginner learning AI prompts for two hours per week. |
Useful Prompt Patterns
Explain
“Explain [topic] to a complete beginner using one analogy and two examples.”
Compare
“Compare [option A] and [option B] in a table using cost, difficulty, benefits and risks.”
Improve
“Rewrite this text for clarity while preserving all verified facts.”
Plan
“Create a step-by-step plan with milestones, deadlines and risks.”
Critique
“Review this draft and identify weak logic, missing information and unclear sections.”
Transform
“Turn this article into a newsletter, five social posts and a short video script.”
Prompt Examples for Common Tasks
Writing
“Act as a professional editor. Rewrite this paragraph for clarity and readability. Keep the meaning, remove repetition and explain every major change.”
Research
“Create a research plan for this topic. List the main questions, possible source types and claims that require verification.”
Freelancing
“Draft a concise project proposal for a client who needs website copy. Include understanding of the problem, deliverables, timeline and three questions.”
Blogging
“Create an SEO-friendly outline for a beginner article on [topic]. Match informational search intent and include FAQs.”
Business
“Analyze this customer complaint and draft a calm response that acknowledges the issue, explains the next step and avoids admitting facts that have not been verified.”
Learning
“Teach me [topic] in four stages. After each stage, ask three questions before continuing.”
Using Examples to Guide AI
Examples can show the desired pattern. A small set of input-and-output examples is commonly called few-shot prompting. Label the examples clearly so their content is not confused with your instructions.
You can provide:
- A sample headline you like.
- A paragraph showing the desired tone.
- A completed table demonstrating the structure.
- A correct and incorrect example.
Use fictional or authorized non-sensitive examples. Confirm permission and approved data handling before using real client or workplace information. Removing a name alone may not remove confidentiality or privacy risks.
Improve Results With Follow-Up Prompts
The first response is often a draft. Use follow-up instructions to refine it.
- “Make this more specific.”
- “Add one practical example to each section.”
- “Identify claims that require verification.”
- “Remove repetition and shorten by 25%.”
- “Rewrite this for a beginner audience.”
- “What important perspective is missing?”
Strong prompting is usually a conversation, not one perfect command.
Beginner-Friendly Advanced Techniques
Ask for a Plan First
Request an outline before asking for the full output.
Break Large Tasks Into Stages
Separate research, structure, drafting and editing when that makes review easier.
Use Evaluation Criteria
Define accuracy, relevance, completeness and format checks before testing.
Request Alternatives
Ask for three options instead of one.
Ask the AI to Identify Uncertainty
Request a list of assumptions and facts needing verification.
Common Prompting Mistakes
Being Too Vague
The AI cannot reliably infer every unstated requirement.
Giving Conflicting Instructions
For example, asking for both maximum detail and extreme brevity.
Providing Too Much Irrelevant Context
Extra information can distract from the task.
Accepting the First Response
Most outputs improve through review and follow-up prompts.
Requesting Guaranteed Accuracy
A prompt cannot remove the need for verification.
Sharing Private Information
Prompts should not contain confidential data without authorization.
A Reliable Prompt Improvement Workflow
Write the Goal
Describe the final result in one sentence.
Add Context
Explain the audience and purpose.
Set Requirements
Define quality, length and limitations.
Generate a Draft
Ask for the first version.
Review Critically
Check facts, relevance and missing information.
Refine and Verify
Use follow-ups and authoritative sources.
Illustrative Mini Case Studies
These fictional examples illustrate possible outcomes, not verified client results or guarantees.
Example 1: Better Blog Outline
A blogger replaces “Give me blog ideas” with a detailed audience, niche and search-intent prompt. The ideas become more relevant and usable.
Example 2: Client Proposal
A freelancer gives the project requirements, client problem and desired tone. The AI drafts a proposal that requires fewer edits.
Example 3: Conflicting Prompt
A user asks for an answer that is both extremely detailed and under fifty words. The result is incomplete because the instructions conflict.
Treat Prompting as Clear Communication
Prompt writing improves when you focus on clarity instead of tricks. Explain the task as though you were briefing a capable assistant who does not know your situation.
Related Lessons and Recommended Resources
Continue Learning on MoneyOnliners
Official External Resources
Your Weekly Challenge
Test Five Prompt Pairs
1. Write five weak prompts you might normally use.
2. Improve each with a clear task, relevant context, requirements and format; add a role only if useful.
3. Test both versions and compare the results.
4. Use follow-up questions to improve the stronger answer.
5. Save your three best prompts in a prompt library.
Practical Prompt Tests
Exercise 1: Improve a vague request
Weak prompt: “Write a customer email.”
Stronger prompt: “Draft a polite reply to this fictional customer question: ‘Can I return my notebook?’ Use only this policy: returns are accepted within 14 days with a receipt, and staff must approve refunds. Do not promise a refund. Keep the reply under 90 words and end with one question about missing information.”
Review whether the response preserves the policy, avoids invented facts, answers the question and fits the tone. A more detailed prompt is only better if the result meets your criteria.
Exercise 2: Separate instructions from source material
Put your task above the text you want processed and mark the source clearly, for example with “SOURCE TEXT START” and “SOURCE TEXT END.” State that the marked text is material to summarize, not instructions to change the task.
Practise with a fictional document. If it contains a request to reveal private information or take an unrelated action, treat that request as untrusted source content. Clear labels help organize a prompt, but do not guarantee protection from prompt injection. Keep consequential actions subject to appropriate review.
Exercise 3: Compare prompts fairly
Run both versions on the same safe task using the same tool and available settings. Use separate new conversations where practical so the first test does not supply context to the second. Record any relevant personalization or supplied files that could influence the result.
Try several representative inputs, not just the easiest example. Change one instruction at a time when investigating a problem. Keep the version that works better across your examples, and record where it still fails.
Your Output Review Scorecard
| Check | Question | Action if It Fails |
|---|---|---|
| Accuracy | Does each factual claim match the brief or verified evidence? | Correct or remove unsupported claims |
| Relevance | Does the answer complete the actual task? | Clarify the task and missing context |
| Format | Does it follow the requested structure and length? | Request a targeted revision and validate again |
| Responsible use | Does it avoid private data, invented promises and unauthorized actions? | Stop, revise the input or seek the required approval |
Save a Useful Prompt Record
Store the prompt’s purpose, safe example input, version date, tool used, review criteria, sample output and known limitations. Keep private information out of a reusable public template. Re-test when you change the prompt, model, source material or workflow.
Ask for Useful Explanations
Request a concise explanation of the result, key assumptions and evidence you can check. For example: “List your assumptions and show the calculation used for the total.” Detailed-looking reasoning does not establish that the answer is correct.
Knowledge Check
Answer first, then open each explanation.
1. Does every prompt need all five parts?
No. Use the checklist flexibly. A simple instruction can be sufficient for a simple task; add details when they improve the result.
2. Can “act as an expert” guarantee accuracy?
No. A role can guide style or perspective, but does not supply qualifications or replace independent review.
3. Does requesting citations verify a claim?
No. Open the original sources, confirm they exist and check that they support the specific statement.
4. Why compare prompts across several examples?
A prompt can work on one easy input and fail on another. Representative tests reveal missing instructions and recurring errors.
5. Does labeling source text make every embedded instruction safe?
No. Labels clarify the task but are not a complete security control. Treat outside content as untrusted and review consequential actions.
Reflection Questions
- Which part of the five-part prompt formula do you usually forget?
- Why does context improve AI output?
- When should you provide examples?
- Why is the first AI response usually a draft?
- How will you organize your best prompts?
Download the Lesson 10 Workbook
Use the Lesson 10 workbook to practise prompt writing and record improvements. Keep your tested prompt pairs, output checks and final revisions from this lesson alongside it.
📘 Download Lesson 10 WorkbookFrequently Asked Questions About AI Prompts
Review the main ideas before continuing to Lesson 11.
What is an AI prompt?
A prompt is the question, instruction or information given to an AI system.
Do prompts need to be long?
No. They need enough relevant detail for the task.
What makes a prompt effective?
A clear task, relevant context, realistic requirements and a usable format. A role is optional.
Can a good prompt guarantee accuracy?
No. Important claims still require verification.
Should I save my prompts?
Yes. A prompt library saves time and improves consistency.
What should I learn next?
Continue to Lesson 11, AI for Writing and Content Creation.