Meta's scaling breaks new winners - any fixes?

ctrjunkie

Standard setup: test campaign, one creative per ad set to validate, then move the winners to a scale campaign after a week or so. That handoff is where everything goes wrong.

I've tried three approaches, and they all have the same problem:

Multiple winners in one ad set - Meta picks an early favourite (the one with more history) and starves the rest, even if their ROAS is better. One winner grabs 80% of spend, others get crumbs.
One winner per ad set - solves starvation but now I'm managing way more ad sets, each going through its own learning phase. I lose Meta's intra-day budget shifting.
Just bump budgets in the test campaign - cleaner but pollutes my testing structure, and scale exclusions don't apply.

So the trade-off is: consolidate and trust Meta (accept starvation) vs. isolate and enforce fair budgets (lose algorithmic optimisation).

A guy in the thread pointed out the Breakdown Effect - that's the technical reason. Someone else said they launch 6-10 ads per ad set in a CBO and keep adding new ad sets rather than messing with existing ones. They claim ad-set level starvation isn't as bad as ad-level. But I still worry about new ads getting buried once winners emerge.

Questions: 1. How do you actually handle this in practice? 2. Anyone using automated rules to cap dominant spend? Does it work or just cause pause/unpause cycles? 3. Does the 20-creatives-per-ad-set Andromeda thing actually work, or only at high spend?

Not looking for 'trust the algorithm' - I tried that and it destroyed my ROAS. Looking for structural solutions.

coopplayer

Meta's algorithm is like that one artist who only paints the same three canvases, ignoring the new masterpiece right in front of it. The Breakdown Effect kills scaling potential for fresh winners-seen it happen a dozen times. My go‑to: launch 6-10 new ads at once into brand new ad sets inside a CBO. Tried dropping them into existing sets, but at scale, Meta just starves the newcomers. Keeps the creative pipeline breathing.

ctrjunkie

Great shout on the Breakdown Effect - that's the exact term for it, and honestly I'd never quite pinned it down like that.

Quick question though: when you run 6-10 ads in new ad sets inside a CBO, does that only work if you launch the whole batch at the same time? Or do you ever add fresh ads back into that same campaign later? My worry is that once 2-3 ads become clear winners, they'll just starve anything new you throw in - same Breakdown Effect, just at ad-set level this time.

Are you constantly cycling new batches into separate campaigns? Or do you eventually merge everything into one ongoing CBO once the winners are clear?

coopplayer

I've found that with CBOs, I usually drop in fresh ad sets containing 6-10 ads each. Once an ad set is live, I leave it completely alone-no tweaking, no pausing. The trick is to keep layering in new ad sets (with new creative) into the same campaign, like a rotating gallery of experiments.

The worry that once a couple of winners surface they'll choke off the newer additions through the same breakdown effect at ad-set level-that's a fair observation, because we see it happen at the individual ad level. But in practice it doesn't play out as badly there. Spend distribution at the ad level is far more right-skewed than at the ad-set level inside a CBO, so the ad sets tend to breathe more evenly than the ads inside them.

neonlights

Mate, if your adsets are actually getting spend, where's the "starvation" coming from? Isn't that just one winner hoarding the budget?