When an account that used to scale suddenly stalls, the instinct is almost always to go looking in the wrong place. Teams rebuild audiences, test new bid strategies, or blame algorithm changes, when the actual bottleneck is usually sitting in the ad library the whole time. Meta's own data attributes 56% of auction outcomes to creative, which means most plateaus are a creative production problem wearing a targeting disguise. 

That was the core diagnosis Jake Abrums, founder of the performance creative agency Odyssey, walked through at Affiliate Summit East this year, laying out exactly why accounts stall and the operating system he uses to unstick them. It's aimed at teams already spending real money and already doing the basics right, who are trying to find the actual lever that's left. 

Why the Account Actually Stalled 

Two structural shifts explain why creative volume, not creative quality alone, is now the ceiling. 

The first is how Meta's delivery logic has changed. Abrums described the old model as a gladiator arena, where accounts built toward one dominant ad that could absorb the whole budget. That model is dead. Andromeda runs more like a matchmaker, spreading spend across a wide portfolio of narrow, specific ads and pairing each with the exact segment it fits. Feed it five variations of the same winning ad and Andromeda will often read them as duplicates, since the system evaluates on-screen similarity, not strategic intent. An account that "stopped scaling" is frequently an account still playing gladiator rules against a matchmaker system. 

The second is that AI has removed the idea bottleneck entirely, which sounds like it should help but often makes plateaus worse. Ask a model to analyze every ad in the account, mine reviews, or pull from competitor libraries and Reddit threads, and it will return more concepts than any team could ever produce. Abrums was blunt about the failure mode here: run that output unfiltered and 70 to 80% of it is unusable. The accounts that recover from a plateau are the ones with someone applying real taste to that output, not the ones simply generating more of it. 

Step 1: Rebuild From Persona Down, Not Campaign Up 

Most stalled accounts are still targeting broad segments with broad messaging. Abrums' fix is to break the audience into micro personas defined narrowly enough that each one implies its own message. Not "moms," but "a first time mom upgrading her kitchen for the safety of her newborn." For each persona, map the motivator, the specific product benefit that answers it, the trigger moment that pushes action, and the stage of awareness they're at. 

Done properly, one working session produces ten personas, five motivators per persona, five matching benefits, and ten trigger moments, which is the strategic backbone for the rest of the system, not a brainstorming exercise you do once and file away. 

Step 2: Stop Treating One Winning Ad as the Finish Line 

A winning message that only exists in one format is a fragile win. Abrums' teams take every proven concept and rebuild it across the format spectrum: static, faceless native footage, fully scripted, street interview style, and elevated cinematic. The point isn't variety for its own sake. Different formats surface to different audience segments inside Andromeda's matching logic, so a message that plateaus in one format can reopen scale in another without any change to the underlying strategy. 

Step 3: Run Four Account Types in Parallel, Not One 

Relying on the brand account alone caps reach before paid spend even enters the picture. Abrums runs four in parallel: the brand account itself, a fourth wall account where a real employee posts in their own voice, creator accounts, and niche publisher accounts unconnected to the brand name, the model behind Odyssey's own sites like The Wellness Review and Men's Performance Digest. He calls this operating like a niche maxing studio, and cites Boardwalk and Frost Buddy as brands running public ad libraries with thousands of live creatives specifically because they treat every niche as its own content surface, not a shared audience to be messaged once. 

Step 4: Recalculate Your Actual Volume Requirement 

A stalled account is frequently just under producing relative to its own spend. Abrums' benchmark: monthly ad budget divided by 2,000 is the minimum viable creative count. A $500,000 monthly budget across Meta and TikTok should be producing at least 250 creatives, roughly 100 core concepts each cut into two to three variations. Teams that plateau are very often running a $500,000 budget on a $100,000 budget's creative output. 

Step 5: Let Organic Absorb the Testing Risk 

Testing everything directly in the ad account is expensive and slow. Abrums' teams push new concepts organically first, since it's free and the signal is real. Across a dataset of over 2,100 organic posts, 11% cleared 2,500 TikTok views with zero spend, 7% cleared 5,000, and 3.5% hit a meaningful estimated media value benchmark, purely organic. Some of the strongest performers came from creator accounts under 12,000 followers pulling views in