AI-generated content briefs that actually rank
The prompt structure we use to brief 40 articles a month without flooding Google with thin content.
AI is excellent at briefs, mediocre at drafts and dangerous at publishing. The leverage is at the brief stage. A well-built brief saves a writer 4 hours of research, lifts ranking probability and stops the team shipping content that all sounds the same.
The prompt structure
Top-10 SERP analysis, common subtopics, missing angles, internal link targets, a 60-word answer block, and a recommended schema set. Feed the model the actual SERP HTML when you can — abstracted summaries lose the specificity that makes briefs useful.
What the human still owns
- Editorial direction and angle.
- Original data, customer quotes, internal benchmarks.
- Opinionated takes — the part that distinguishes your content from the SERP.
- Fact-checking every claim the model surfaces.
Quality bar
If the brief reads like a template, the article will too. Reject any brief where you couldn't guess the publication from the angle. Briefs should feel specific to your brand and your customers — generic briefs are how you end up with a content farm.
Cost reality
A good brief costs 15–25 minutes of prompt iteration plus £0.40 in API calls. A bad brief looks free. Time-to-publish drops 40% with disciplined briefs; quality scores in human review usually rise too.