Not all prompts are created equal...
A poorly written prompt leads to vague, misleading, or irrelevant responses, while a well-crafted one unlocks precise, valuable, and actionable insights.
In this guide, we’ll walk you through a step-by-step process to develop, refine, and optimise prompts using ChatGPT.
Each section will focus on a theoretical use case, showing:
✅ A weak prompt (with an explanation of why it fails)
✅ Two refined prompts (with explanations of how they improve the output)
✅ Key refinements summarised
By the end, you’ll be able to start crafting high-quality prompts that consistently produce accurate and insightful responses. But this is just a starter, the more you prompt, the more you will learn.
Let’s dive in! 🚀
Use Case: Analysing Customer Feedback for Trends
💡 Imagine you have a dataset of customer reviews and you want ChatGPT to help identify key insights, such as pain points and positive themes.
Step 1: Defining the Objective 🎯
Why It Matters
🧐 Ever asked ChatGPT a question and got a vague, wishy-washy answer? That’s a sign of a weak prompt. The clearer you are, the better the response.
A strong prompt clearly specifies what you want ChatGPT to focus on. If the objective is vague, the response will also be vague.
Weak Prompt
🚫 “Tell me about this dataset.”
🔴 Why This is Weak:
- The request is too broad, so ChatGPT won’t know whether to focus on trends, anomalies, or individual data points.
- The output will be generic and may not align with the intended analysis.
Refined Prompt 1
✅ “Summarise key trends in this dataset, focusing on patterns in customer satisfaction scores.”
🔵 Why This is Better:
- It adds specificity by requesting key trends.
- It defines a focus area (customer satisfaction).
Refined Prompt 2
✅ “Analyse this dataset and list three common pain points and three positive themes that appear in customer feedback.”
🔵 Why This is Even Better:
- It asks for structured output (pain points and positive themes).
- It provides a clear expectation for how many insights should be listed.
💡 Key Refinements:
🔹 Removed ambiguity
🔹 Specified a focus area
🔹 Structured the output
Step 2: Identifying Common Prompt Mistakes ⚠️
Why It Matters
🤔 Ever asked a vague question and got a vague answer? That’s because ChatGPT responds to what you ask, not what you meant to ask.
Vague prompts produce vague responses. Identifying common mistakes ensures the AI delivers relevant insights.
Weak Prompt
🚫 “What do customers think?”
🔴 Why This is Weak:
- The question lacks direction. It doesn’t specify what aspect of customer opinion to focus on.
- ChatGPT might provide an overly general response, missing key themes.
Refined Prompt 1
✅ “Summarise customer reviews. Identify common complaints, feature requests, and overall sentiment.”
🔵 Why This is Better:
- It clarifies what insights should be extracted.
- It adds structure by listing specific categories (complaints, requests, sentiment).
Refined Prompt 2
✅ “Analyse customer feedback and categorise responses into:
1️⃣ Pain Points
2️⃣ Positive Features
3️⃣ Suggested Improvements”
🔵 Why This is Even Better:
- It forces clear segmentation into categories.
- It improves readability and structure for easier analysis.
💡 Key Refinements:
🔹 Specified insight categories
🔹 Provided a structured breakdown
Step 3: Structuring Prompts for Precision 🏗️
Why It Matters
🔍 The way you structure your question directly affects the way ChatGPT structures its response.
Well-structured prompts improve clarity, ensuring ChatGPT delivers focused and usable responses.
Weak Prompt
🚫 “Explain this dataset.”
🔴 Why This is Weak:
- It doesn’t define what part of the dataset should be explained.
- The response could be too general or overly detailed without direction.
Refined Prompt 1
✅ “Identify the most common themes in this dataset. Summarise findings in under 50 words.”
🔵 Why This is Better:
- It prioritises key themes rather than random details.
- It sets a word limit to ensure concise output.
Refined Prompt 2
✅ “Analyse the dataset using the following structure:
📌 Customer Demographics – Age and location trends
📌 Product Preferences – Most frequently mentioned features
📌 Customer Sentiment – Positive vs. negative feedback”
🔵 Why This is Even Better:
- It provides a step-by-step breakdown for a well-organised response.
- It ensures consistency in AI-generated insights.
💡 Key Refinements:
🔹 Improved specificity
🔹 Added structure
🔹 Set a word limit for clarity
Step 5: Troubleshooting and Debugging Prompt Issues 🛠️
Why It Matters
🔧 Even the best prompts sometimes fail. Knowing how to tweak them will save you time and frustration.
Even well-structured prompts sometimes fail. Knowing how to debug them helps refine results.
Common Issues and Fixes
Issue: The response is too generic
🚫 “What are the main takeaways from this text?”
✅ Fix: “Summarise the key takeaways with three specific examples from the text.”
Issue: The response lacks key details
🚫 “Give me marketing strategies.”
✅ Fix: “Provide five marketing strategies for increasing customer retention. Each must include a real-world example.”
Issue: The response is too long or disorganised
🚫 “Write a detailed analysis of this dataset.”
✅ Fix: “Summarise the dataset analysis in three sections: Trends, Anomalies, and Key Insights. Keep each section under 50 words.”
💡 Final Tip:
Even expert prompt engineers don’t get it right the first time. Testing and tweaking is part of the process, so don’t be afraid to experiment!
Final Thought: Prompt Development is an Iterative Process 🚀
A great prompt = great results. Think of prompt engineering as crafting the perfect question rather than just issuing a command.
✅ Be clear and structured
✅ Refine and test your prompts
✅ Experiment with constraints and role-based instructions
🔥 A great prompt is like a backstage pass. It gives you access to deeper, more valuable insights. Ask better, and you’ll get better!
This training guide is yours to use however you like. Share it, apply it in your business, or adapt it to fit your brand. That is what DIY culture is all about! If you find it valuable, consider supporting my work with a coffee or a beer through my Buy Me a Coffee link below. Cheers! ☕🍻
