How to Write Better AI Prompts: Step-by-Step Guide with Examples

Most people get disappointing results from artificial intelligence because they treat AI tools like simple search engines instead of intelligent business collaborators. If your prompt is vague, your resulting output will be generic.

Writing effective prompts is a systematic, repeatable skill. In this practical guide, we share five step-by-step techniques to immediately improve your prompts across ChatGPT, Claude, and Google Gemini.

---

1. Apply the C-T-R-O Prompt Formula

Before clicking send, check if your instruction contains the four essential components of the C-T-R-O Formula:

• C for Context: What background information does the AI need to understand your situation?
• T for Task: What exact action should the AI execute?
• R for Role: Who should the AI act as during the task?
• O for Output Format: How should the response be formatted (table, bullet points, JSON)?

---

2. Weak Prompt vs Strong Prompt Examples

Example 1: Email Marketing Campaign

• ❌ Weak Prompt: “Write an email about our new website design course.”
• ✅ Strong Prompt: “Act as a senior email conversion strategist. Write a three-email sales sequence targeting university graduates who want to learn website design using WordPress. Tone: encouraging, professional, and persuasive. Include a strong call-to-action link to our Lagos campus page.”

Example 2: Market Analysis

• ❌ Weak Prompt: “Tell me about AI in Nigeria.”
• ✅ Strong Prompt: “Act as a market researcher. Summarize five ways corporate banks in Lagos are applying AI for fraud detection in 2026. Present your answer in a three-column table: AI Application, Key Business Benefit, Main Security Challenge.”

---

3. Instruct the AI to Ask You Clarifying Questions

One secret technique used by expert prompt engineers is forcing the AI to ask questions before generating its final answer.

Add this single line to any complex prompt:

“If any background details are missing or ambiguous, ask me three clarifying questions before drafting your final response.”

This simple addition prevents assumptions and ensures the output matches your exact requirements.

---

4. Provide Few-Shot Examples (Pattern Matching)

Large language models excel at matching patterns. By providing one or two ideal input and output examples directly inside your prompt, you guarantee that the model adopts your preferred structure and phrasing style.

---

5. Master Iteration and Refinement

Never view the first response as final. Use precise follow-up prompts to refine your output:

• “Make the introduction more compelling for executive readers.”
• “Rephrase paragraph two to maintain a warmer, more engaging tone.”
• “Simplify the technical terms so a non-technical manager can understand.”

---

Learn Hands-On Prompt Engineering in Lagos

Want expert feedback on your prompts from experienced corporate mentors?

Join our four-week physical Prompt and Context Engineering Course at TCS Plus Academy or explore our Generative AI for Work Course.