What Good Data work Prompts Have in Common

You don't need a hundred data work prompts. You need five you trust and actually reuse.

Anchor to one reader

Treat the first output as a draft, not a verdict. The prompt does the typing; you keep the judgment about what ships.

Start with the smallest version

Ask for the reasoning, then the result. When the model shows its steps you can see exactly where to push back.

• Keep one ask per prompt — chaining beats cramming
• Compare two outputs instead of trusting one
• Have the model flag what it could not cover

Write down what you never want as explicitly as what you do. The exclusions often do more work than the instructions.