Creative fatigue is real and has gotten worse. A winning ad used to have 8–12 weeks of useful life on Meta. Now it's 3–5 weeks at scale, faster on TikTok. You can't produce your way out of that with a traditional studio pipeline — the math doesn't work.
So here's the workflow we use. I want to be upfront: it is not elegant, it doesn't always work, and we rework parts of it every quarter. But it beats what most in-house teams are running.
Think in hooks, not ads
An "ad" is the finished thing. A "hook" is the first 2–3 seconds: the sentence, visual, or motion that makes someone stop scrolling. 90% of performance is decided in the hook. The rest of the ad mostly exists to not lose the people the hook caught.
So when we brief creators, we don't ask for ads. We ask for 6 hooks on a concept, each 4–8 seconds long. We stitch them to a common middle and end. One 90-minute shoot yields 12–18 testable variants that way.
Creators beat studios for this kind of work
Controversial with creative directors, but: briefed creator content outperforms polished brand content on paid social about 3 times out of 4 in our tests. It's cheaper ($300–600 per deliverable vs. $5K+ for studio), faster (5-day turnaround), and looks native to the feed.
The hard part isn't finding creators. It's briefing them well. A bad brief produces bad content no matter who shoots it. Our briefs include the hook we want, a reference video showing the vibe, the product claim we can and can't make (legal stuff matters), and a hard no-list.
We keep a running panel of 15–25 creators per client and rotate them. No single creator's face should carry more than 25% of active creative — if they burn out on one account they're done for the brand.
The weekly cadence
Roughly — we flex this all the time based on account size and how things are performing:
- Monday: pull last week's results. Kill anything underperforming by 30%+ vs account CPA. Scale things that are beating CPA by 30%+.
- Tuesday: brief next week's batch. Usually 2–3 concepts × 4 hooks.
- Wednesday–Thursday: receive footage, flag anything unusable.
- Friday: edit, upload, launch into the test campaign with small budgets.
You'll notice this doesn't give you time to be precious about anything. That's intentional. Most variants won't win. Volume is the point.
What I still get wrong
We over-invest in hooks that worked 6 months ago and keep trying to resurrect them. The audience has moved on. I know this and still do it. I think most paid social teams do.
The other thing — we sometimes test too many variants too fast and don't give any of them enough data to judge fairly. You need at least $300–500 in spend per variant on a mid-ACV product before the numbers mean anything. Below that you're reading noise.