Hernan VazquezFounder and CEO, Rapid Scale Framework · Creator of The Rapid Scale Framework, Author of Low Ticket Profits

Why Meta Wants 30 Creatives In Your Ad Set, Not 5 #shorts

23 Sept 2026 · 1 min

Watch on YouTube

1:00

In this YouTube short, the speaker explains that Meta’s ad‑delivery system, now called Andromeda, no longer works with the old playbook of testing a handful of creatives, picking a winner, and scaling it. The algorithm now thrives on variety: advertisers should load each ad set with 20‑30 different creatives that span multiple formats, aspect ratios, and messaging angles. The system will serve the most suitable creative to each individual viewer—some will see a square image, others a vertical video, others a testimonial screenshot. Because most creatives will receive little spend, that’s acceptable; the breadth of options lets the algorithm find the best match. The recommended workflow is to launch a fresh batch of 20‑30 varied creatives each week, recycle those that didn’t get enough exposure, scale the ones that perform, and discard the clear under‑performers. The shift moves away from hand‑picking a single “perfect” ad toward giving the algorithm enough diversity to optimize per‑user delivery.

Chapters

Meta's algorithm changed and a lot of advertisers are still using the old playbook. It used to be enough to test a few creatives, find the winner, and scale it. That's not really how Andromeda works now. The system wants a lot more variety. Instead of giving it 3 to 5 creatives, you want to give it 20 to 30 in a single ad set with different formats, different aspect ratios, and different messages. Most of them won't get much spend, and that's fine. The algorithm is picking the best fit for each person who sees the ad. One person might get a square image, someone else might get a vertical video, and someone else might get a testimonial screenshot. So if you only give it a few options, you're limiting what it can do. What's working now is loading each ad set with a wide mix of creatives, then launching a fresh batch every week instead of constantly adding to the old one. You recycle what didn't get a fair shot, scale what works, and cut what clearly doesn't. That's the shift. You're not trying to handpick one perfect ad anymore. You're giving the algorithm enough variety to find the right match.