Decide whether a press release is worth writing, then write it
Use before announcing something, when you are not sure it is actually news.
Fill in before running
Replace each placeholder with your own detail. The more specific you are, the less the model invents.
- {{WHAT_WE_ARE_ANNOUNCING}}
- {{OUR_REASONING}}
- {{AFFECTED_PARTIES}}
- {{FACTS_FIGURES_DATES}}
- {{SPOKESPERSON_NAME_AND_TITLE}}
Getting a better result
- Run step one on its own first; most announcements score below three and the release is wasted effort.
- If the quote reads like something no human would say aloud, ask for it again with the instruction to write it as spoken answer to a question.
- NEEDS FIGURE markers must all be resolved before anything is sent - do not let a draft go out with them in.
Questions about this prompt
When should I run this rather than just writing the release?
Before anything is drafted, when someone internally has decided an announcement is news. Step one scores it on whether anything changes for people outside the company, so run that alone first. Whether press releases still earn coverage at all is contested, and a score below three usually points at a blog post.
What do I need before step one can score anything?
The announcement, who it affects outside your company, and FACTS_FIGURES_DATES, since the score depends on there being a number and a date. Also the spokesperson name and title. If AFFECTED_PARTIES turns out to be only your own staff and customers, expect a low score, which is the answer you needed.
What comes back if the announcement is news?
A newsworthiness score with reasoning and a plain recommendation, then a release only when the score reaches three. The verdict is the valuable half. Where it does write the release, read the spokesperson quote closely, since the prompt is barred from the delighted, excited and proud formulations that make quotes unusable.
What is the error that shows up in the sent copy?
Letting a draft go out with NEEDS FIGURE markers still in it. The prompt writes them rather than estimating, which is deliberate, and one unresolved marker is the detail a journalist notices first. Resolve every one before distribution, or cut the sentence that needed the number.