.png){kind=logo}
.png)
Intro
Every Brand running Creator ads on Meta eventually hits the same wall: the winning ad that performed for weeks suddenly stops working, and nobody can say why or what to do next.
Scaling short-form Creator ads still feels like guesswork: when to kill a test, when a winner is truly exhausted, and how many new creative assets you actually need.
So we analyzed $130M in ad spend across 65,000 Meta Partnership Ads - one of the largest datasets of its kind - to replace that guesswork with data. The result is a first-of-its-kind report on building a Creator-led creative engine that beats fatigue and never runs out of winning ads.
Partnership Ads are exploding, but they're perishable
Partnership Ads are Meta's fastest-growing ad product. As Mark Zuckerberg put it on Meta's Q1 2026 earnings call: "Brands [are] increasingly turning to Creators to promote their products. This is contributing to rapid growth in our Partnership Ads product, with its revenue run rate more than doubling year over year in Q1 to $10 billion."
The format works because the ad runs from the Creator's handle with the paid-partnership label - it reads as a person recommending something, not a Brand advertising. And Meta optimizes it against two datasets at once: the Creator's follower graph and engagement history, plus the Brand's pixel.
But the data shows most Brands squander that advantage in three ways: they exhaust winners too fast, kill promising tests too early, and retire creative that still has audience left to reach.
What the Agentio data shows
1. Drop traditional UGCs and run Meta Partnership Ads instead. Head-to-head, Partnership Ads deliver +19% CTR, +10% CVR, and −5% CPA compared to traditional licensed UGC run from the Brand handle. The Creator's identity and audience signal drive the gap - run the same asset from the Brand account and the trust signal is gone. The Creator's identity and audience signal drive the gap. Run the same asset from the Brand account, and the trust signal is gone
2. Fund every ad to ~$1K and launch it in a clean ad set. Only 1 in 5 tested ads becomes a winner. 45% of eventual winners still look like losers when performance is assessed before reaching $100 in spend. Set a ~$1,000 spend floor before the kill call, and launch tests into ad sets with no incumbent ad so budget can actually reach them.
3. Build a monthly testing engine. Winners fatigue after 36 days on average, and when kept alive past that point, median CPA increases 1.4x (1.8x on the spend-weighted average). Maintaining 10 concurrent winners takes ~40 new ad tests per month - replacement is a schedule, not a reaction.
4. Redeploy fatigued winners to YouTube Shorts. 34% of US YouTube Shorts users aren't on TikTok or Instagram Reels, and CPMs for YouTube Creator Partnership Boosts run 30% lower than Meta Partnership Ads. A fatigued winner isn't dead - it just needs a fresh audience.
Want to build a creative engine that never runs out of winners?
The full report includes:
- Placement-by-placement performance deltas: Partnership Ads vs. licensed UGC
- Hit-rate benchmarks across 137 Brands, and the spend floor every test deserves
- The fatigue clock: when winners decay, and what it costs to keep them alive
- The replacement math: exactly how many ads to test monthly at your spend level
- The YouTube Shorts arbitrage, and how to capture it before it closes
- Full methodology and glossary
