Search

The Prompt That Changed Everything

Michael Lee, MBA
2 days ago
3 min read

You’re in the middle of drafting a performance review. You open ChatGPT and type:

"Can you help me write a review for my team?"

The response? A bland, overly generic blurb that sounds like it came from a corporate template.

Now, try this instead:

"You are a HR Manager in a logistics company. Help me write a professional and motivating performance review for a high-performing junior analyst named Sarah, who improved delivery time by 25% and took initiative in streamlining warehouse workflows. The tone should be encouraging but also suggest areas for future growth."

Suddenly, the output is specific, nuanced, and actually useful.

That’s the difference a PERFECT Prompt makes.

🤖 Why Prompting Isn’t Just Fluff

There’s a myth that AI just “gets it.” But in reality, LLMs (Large Language Models) work best when your instructions are rich in context and clarity. Like briefing a new hire, the quality of what you get back depends on what you put in.

In a world where AI tools are writing marketing emails, code, reports—even making business decisions—it’s time we treat prompt writing like the skill it is.

🧠 Introducing the PERFECT Prompt Framework

We developed the PERFECT Prompt to help teams go from vague input → valuable output. Here's the breakdown:

Purpose – What task are you trying to accomplish?
Expectations – What should the output look like? (e.g., format, length)
Role – Who is the AI pretending to be? (e.g., HR Manager, Data Analyst)
Focus – Key data, themes, or examples the AI should emphasize
Emotions – Tone or mood of the content
Constraints – Any rules to follow (e.g., avoid jargon, 3 bullet points max)
Target – Who is the audience for the output?

This structure doesn't overcomplicate—it guides your prompt like GPS.

📊 Real Business Examples

Here’s how companies are already reaping the benefits of structured prompting:

💳 Klarna (Retail)

Use Case: Customer service chatbots
Prompt Style: “You are a support agent at Klarna. A customer claims a refund was not received after 14 days. Respond politely, reassure them, and offer next steps.”
Result: Faster issue resolution and more natural, brand-aligned replies.

🔗 Zapier (SaaS Automation)

Use Case: Generating custom automation help based on user context
Prompt Style: “You are a Zapier assistant. A user is trying to integrate Shopify with Google Sheets to track inventory. Suggest three automation workflows they can try.”
Result: Higher customer satisfaction and engagement

🧠 Morgan Stanley (Finance)

Use Case: Internal knowledge assistant for 16,000 advisors
Prompt Style: Custom-tuned to deliver compliance-aligned financial advice and internal process answers based on firm data.
Result: Reduced search time and higher productivity.

⚖️ “Before & After” Prompting in Action

Let’s look at a typical example:

✍️ Unstructured Prompt:

"Write a LinkedIn post about AI in sales."

Output:

“AI is changing the way we sell. It helps automate tasks and understand customers. Embrace the future of AI in sales!”

✅ Accurate, ❌ Uninspired.

🎯 PERFECT Prompt:

“You are a sales leader at a B2B tech startup. Draft a LinkedIn post (under 150 words) celebrating how AI tools like Gong and ChatGPT helped your SDR team reduce email response time by 40% and increase SQLs. Keep it upbeat, with a tone of optimism.”

Output:

“In just three months, we saw a 40% drop in response time and a big boost in sales-qualified leads—all thanks to smarter workflows powered by Gong and ChatGPT. AI isn’t replacing our team. It’s supercharging them. Onwards! 🚀”

✅ On-brand, engaging, and useful.