The trap with data work prompts is asking for the finished thing. Ask for options instead, then bring your taste to the edit.
Make the output reviewable
Ask for the reasoning, then the result. When the model shows its steps you can see exactly where to push back.
- Name the audience and the single goal before you generate
- Lower the reading level, then check what broke
- Compare two outputs instead of trusting one
Cut before you polish
Lock the parts that never change and leave one slot for the variable. Now every run shares a family resemblance instead of starting from scratch.
Lock the parts that never change and leave one slot for the variable. Now every run shares a family resemblance instead of starting from scratch.