Automated QA for 400+ ad variations, without hiring a single contractor
The 'middle manager' prompt pattern that catches off-brand creative before it ever ships, and how to wire it up.
Brand teams shipping 400+ creative variations a quarter have a quiet problem: human QA doesn't scale, but a single off-brand asset in market is worse than three approved-and-mediocre ones. Most teams either over-staff QA or under-ship volume. There's a third option.
The 'middle manager' pattern uses a single Claude or GPT-4 evaluator with three things wired into its system prompt: the brand voice charter (verbatim), the do-not-say list (verbatim), and three reference assets that are explicitly on-brand and three that are explicitly off-brand. The evaluator returns a pass/flag/fail with a one-sentence rationale.
Expect the evaluator to clear most variations and flag a minority for human eyes, and expect some of those flags to be over-cautious. Measure your own pass rate before you trust it; we have not run a documented pilot and will not quote one.
The deployment lift is small: a system prompt, a do-not-say list, and a handful of reference assets. The budget argument is that QA capacity gets rerouted to higher-leverage work rather than added.
The prompt template, the do-not-say list format, and the reference-asset selection criteria are all in the appendix.
The model isn't the creative director. It's the middle manager who reads the brief twice and says 'this isn't quite right' before the work hits the CD's desk.
- 01Brand voice charter + do-not-say list + reference assets = working QA evaluator.
- 02Measure your own pass rate before trusting the evaluator unattended.
- 03Evaluator runs are cheap relative to a QA contractor row; price your own usage.
- 04Reroute the saved contractor capacity, don't just delete it.
- Published
- Jun 04, 2026
What changed. What it means. What to do next. Two briefings each week.
Two briefings a week. Unsubscribe anytime.
- Issue 12 · Tactics · Jun 14, 2026The 4-prompt chain that exposes any competitor's creative pattern in 9 minutes
- Issue 08 · Tactics · May 31, 2026The 200-document RAG pipeline that lets any model write in your brand voice
- Issue 04 · Tactics · May 17, 2026Voice search is back, and podcast feeds are the unexpected SEO surface