ChatGPT Prompt Engineering Tips — Write Better Prompts in 5 Minutes
- Five practical prompt structures that actually work
- How to get consistently better outputs without guessing
- Common mistakes that make ChatGPT give generic answers
I spent months writing vague prompts and wondering why ChatGPT kept giving me bland, Wikipedia-style answers. Then I started paying attention to the structure of my prompts, and the difference was honestly night and day. Here’s what I’ve learned after sending thousands of prompts over the past year.
Why Most People Get Mediocre Results
The problem isn’t ChatGPT — it’s how we talk to it. Most people type something like “write me an email” and expect magic. But ChatGPT doesn’t know your tone, your audience, or what “good” looks like to you. You have to give it context, just like you would when briefing a coworker.
Think of it this way: if you asked a new intern to “write something about marketing,” you’d get something generic. But if you said “write a 200-word LinkedIn post targeting SaaS founders about why cold email still works, using a conversational tone” — that’s a completely different output.
Five Prompt Structures That Work Every Time
Start with “You are a [role]” to prime the model. Then state the task clearly, and specify the output format. Example: “You are a senior copywriter. Write 5 subject lines for a B2B SaaS cold email. Format: numbered list with a brief explanation of why each works.”
Give ChatGPT a bad example and ask it to improve it. This anchors the model and produces sharper results. “Here’s my current landing page headline: ‘We Offer Solutions’. Rewrite it to be specific, benefit-driven, and under 10 words.”
Add boundaries. “Explain quantum computing to a 12-year-old. No jargon. Maximum 150 words. Use one analogy.” Constraints force creativity and prevent generic filler.
For complex problems, ask ChatGPT to “think step by step.” This dramatically improves reasoning accuracy. Instead of “What price should I set?”, try “Walk me through the pricing factors for a SaaS product targeting small businesses. Consider costs, competitor pricing, and perceived value. Then recommend a price range.”
Don’t try to get the answer in one shot. Start broad, then narrow. First prompt: “Draft an outline for a blog about remote work.” Second: “Expand section 3 with statistics.” Third: “Make the tone more conversational.” This is how professionals actually use it.
Quick Reference: Prompt Patterns
| Pattern | When to Use | Example Opener |
|---|---|---|
| Role + Task + Format | Writing tasks, content creation | “You are a [role]. Write [task] in [format].” |
| Before & After | Editing, improving drafts | “Here’s my current version… improve it by…” |
| Constraint Method | Need focused, concise output | “Explain X in under Y words, using…” |
| Chain of Thought | Analysis, reasoning, decisions | “Think step by step about…” |
| Iterative Refinement | Complex projects | “Start with an outline, then we’ll refine.” |
Three Mistakes to Avoid
Mistake #1: Being too vague. “Write something about AI” will give you something about AI. Something generic. Always include audience, purpose, and tone.
Mistake #2: Ignoring the system message. If you’re using the API or custom GPTs, the system message is where you set the personality and ground rules. Most people leave it empty.
Mistake #3: Never iterating. The first output is a draft, not a final product. Treat ChatGPT like a brainstorming partner, not a vending machine.
I used to write prompts like “help me write an email.” Now I write “You’re a customer success manager at a SaaS startup. Write a re-engagement email to a customer who hasn’t logged in for 30 days. Tone: friendly but urgent. Include a specific CTA. Under 150 words.” The second approach saves me 10+ minutes of back-and-forth editing. The initial prompt takes 30 seconds longer to write, but the output is usable on the first try about 80% of the time.