Prompt Engineering: Why Some Prompts Work Like Magic — And Others Flop
- Rebecca Thachil
- Sep 8
- 2 min read
Updated: Sep 11
If you’ve ever used ChatGPT and thought, “That’s not what I meant,” chances are it wasn’t the AI — it was the prompt.
Prompt engineering is quickly becoming the skill that separates casual AI users from power users. And the difference between a good and bad prompt? Night and day.
So here are 3 solid examples of great prompt engineering — and 3 that totally miss the mark.
3 Great Prompt Engineering Examples
1. The “Role + Goal + Tone” Prompt
“You are a B2B marketing expert. Write a LinkedIn post summarizing this blog in 3 punchy points, using a witty and confident tone.”
Why it works: It sets the role, clarifies the task, and gives a tone. The AI knows who it’s pretending to be, what it’s supposed to do, and how to sound.
2. The Step-by-Step Instruction Prompt
“Explain this concept to me like I’m five. Use simple examples and break it down in three steps.”
Why it works: Clear structure + target audience = more focused, easy-to-understand output. Great for learning or simplifying something technical for a customer.
3. The Iterative Builder Prompt
“Give me 3 headline options for this email. I’ll give feedback and ask you to improve the best one.”
Why it works: It turns the AI into a creative partner. You set up a back-and-forth loop that improves results through collaboration — not just a one-and-done.
3 Not-So-Great Prompt Examples
1. The Vague Prompt
“Help me with my marketing.”
Why it fails: Too broad. The AI doesn’t know if you want strategy, copy, campaign ideas, or a performance report. You’ll likely get something generic and not useful.
2. The Confusing Multi-Task Prompt
“Write me a cold email and also tell me if the CRM is broken and also what’s the best channel for this?”
Why it fails: Multiple asks in one breath = chaos. The AI tries to do too much at once and none of it lands well. Break it into separate prompts instead.
3. The “Go Figure It Out” Prompt
“Write me something amazing.”
Why it fails: No direction, no context, and a lot of hope. AI needs something to work with. Without that, you’ll get something bland and unoriginal.
Takeaway: Want Better Output? Be a Better Input-er.
Prompt engineering isn’t about being fancy — it’s about being clear, structured, and intentional. You don’t need to overcomplicate it. You just need to guide the AI like you’d guide a new teammate.
If you can do that? AI becomes your best brainstorming buddy, copy editor, and strategist rolled into one.
Now that you’ve got a handle on AI, here’s what’s next for you: Join our free live session – “Starting a Career in RevOps” to explore real-world roles, skills, and career paths.
Want access to the session recordings, exclusive downloads, and peer discussions? Join the Reklik community — where RevOps learners connect, share, and grow together.
Prompt engineering is just one piece — we unpacked a lot more in our AI Terminology and Techstack webinar inside the Reklik community. Tap Join Reklik Community to get in and be part of the session.
