Give the Style Rules You Actually Want Enforced

A precision habit for editing: give ChatGPT the specific style rules to enforce — voice, banned words, formatting conventions — instead of a vague 'improve this', so edits are consistent and on-standard.

'Make it better' lets the model apply its own generic house style, which drifts from yours and varies turn to turn. A concrete rule set — active voice, no em dashes, British spelling, sentences under 25 words — gives it an objective standard to edit against, so the output matches your voice and stays consistent across documents.

Editing to a defined standard: a publication's house style, a brand voice, a team's documentation conventions, or your own hard preferences.

Early creative drafting where imposing rules too soon flattens exploration — save strict enforcement for the polish pass.

Edit this to my style rules:
- Voice: <e.g. active, direct, second person>
- Always: <conventions>
- Never: <banned words, punctuation, tics>
- Spelling/format: <e.g. British, sentence case headings>

Apply only these; don't impose other 'improvements'. Text: <paste>.

Giving the rules 'no em dashes, no "leverage", British spelling, max 25-word sentences' produces edits that read like your publication instead of generic ChatGPT prose.

• Saying 'improve this' and getting the model's house style, not yours.
• Listing vague goals ('more professional') instead of concrete rules.
• Enforcing strict style during creative drafting and killing the ideas.