You know your prompts are sloppy. You’ve seen the results—generic, off-target, or just plain wrong. The solution isn’t to learn prompt engineering. It’s to think before you type.
Why “Just Ask ChatGPT” Doesn’t Work
AI models are trained on billions of words, but they’re not trained on your specific context. When you type “Help me with my presentation,” the model has to guess:
- Is this a sales pitch or a team update?
- Who’s the audience—executives or peers?
- What format do you want—slides, a script, or bullet points?
- How long should it be?
- What tone should it use?
Every ambiguity is an invitation for the model to guess wrong.
The Five-Step Pre-Prompt Framework
Before you ask an AI for help, answer five questions:
1. Define the goal. What are you trying to achieve? Not “create a presentation,” but “create a 10-slide presentation to persuade the board to approve a £50,000 budget increase.”
2. Define the role. What role should the AI play? A consultant? An editor? A copywriter? Setting the role shapes the output. “Act as a sceptical editor” produces different results than “Act as a supportive coach.”
3. Define the audience. Who will read or hear this? Customers, colleagues, managers, or strangers? The audience determines tone, depth, and structure.
4. Set the tone. Formal, conversational, technical, persuasive? Don’t assume the model will guess correctly. Tell it.
5. Describe the output format. Bullet points, paragraphs, a numbered list, a table? Be specific. If you want three options, say so. If you want each option under 100 words, say that too.
How These Steps Reduce AI Errors
They break down ambiguity. The more specific your input, the less the model has to infer.
They improve specificity. When you define the role, audience, and constraints, the model has a clear target.
They reduce hallucinations. Vague prompts invite invention. Clear prompts invite precision.
Worked Example
Original prompt: “Write a blog post about productivity.”
After applying the framework:
Goal: Create a 600-word blog post that helps freelancers manage their time without productivity tools.
Role: Act as a pragmatic productivity consultant who avoids buzzwords.
Audience: Freelancers aged 25–40 who are sceptical of productivity hype.
Tone: Conversational, direct, no corporate waffle.
Output: Three sections—morning routines, managing distractions, and knowing when to stop. Each section 200 words. Use subheadings.
The difference is night and day. The original prompt could produce anything. The revised prompt produces exactly what you need.
Quick Reference Checklist
Before you prompt:
- Have I stated the goal?
- Have I defined the role?
- Have I described the audience?
- Have I specified the tone?
- Have I described the output format?
If the answer to any question is no, you’re leaving the door open for the model to guess.
PreStep Automates These Steps So You Don’t Have to Remember Them
PreStep asks the right questions, structures your thinking, and produces a clear brief you can feed to any AI. No more guessing. No more sloppy prompts.