The best Meta ads testing structure post-Andromeda is no longer a single universal setup. As of 2026, there are four proven structures that work depending on your budget, your tolerance for testing waste, and how much you want to trust Meta’s AI to allocate spend for you. After spending millions per month on Meta and managing 70+ ecommerce brands, we have stress-tested every combination of ABO (Ad Set Budget Optimization) and CBO (Campaign Budget Optimization), single-campaign and split-campaign, manual bidding and highest volume bidding. This guide breaks down exactly which structure to use, and why the old “test in ABO, scale in CBO” playbook is no longer the default winner.
TL;DR
– There is no one-size-fits-all Meta ads testing structure post-Andromeda. The right choice depends on budget, SKU count, and how much testing waste you can absorb. – ABO testing + CBO scaling still works when you want maximum control and are willing to force spend on creatives Meta would otherwise ignore. – One campaign, one adset, test and scale is now the strongest setup for brands spending under $1K to $2K per day per product. – CBO with multiple adsets lets Meta forecast winners before spending, ideal when you want to minimize wasted testing budget. – Engagement rate ranking is the most underused metric for predicting which creatives Meta will actually distribute. – Mixing highest volume, lowest cost bidding with cost cap or bid cap inside the same campaign breaks performance. Pick one.
Why Post-Andromeda Changed Meta Ads Testing Forever
Post-Andromeda testing rewards consolidation, not fragmentation. After Meta rolled out the Andromeda algorithm update, the platform got significantly better at understanding who to show your creatives to, with learning happening at the ad level and being pushed up to the adset level. That shift broke a lot of legacy testing structures.
The old playbook was: build one ABO testing campaign with five to ten adsets, find a winner, duplicate it into a CBO scaling campaign. That still works, but it now creates two structural problems:
1. Audience overlap across separate testing and scaling campaigns inflates CPMs. 2. Learning gets fragmented across multiple adsets and campaigns, which weakens Meta’s ability to optimize at the account level.
Meta is actively pushing advertisers to consolidate ads inside fewer adsets, because that gives the algorithm more data points per adset to refine targeting and distribution. For the deeper context on what changed, our complete guide to the Meta Andromeda update walks through the full attribution and distribution shift.
For reference, Meta’s own documentation on Advantage+ campaign budget confirms that consolidating budget at the campaign level helps the system reallocate spend toward higher-value opportunities in real time.
The Two Big Decisions Before You Pick a Structure
Before choosing between ABO and CBO, answer two questions honestly.
Question 1: Are you willing to waste budget on testing?
If yes, ABO is for you. You force Meta to spend on every creative angle, persona, and concept you launch, even the ones the algorithm would naturally ignore. You learn more, but you spend more to learn.
If no, CBO is for you. Meta’s AI runs a forecast on every ad before it spends a dollar, then allocates budget toward the creatives it predicts will perform best. You waste less, but you let Meta decide what gets tested.
Question 2: Do you trust Meta to pick winners for you?
Meta is right roughly 90% of the time, in our experience across millions in monthly spend. But the algorithm balances two goals: advertiser performance and platform user experience. That means a creative with strong engagement rate ranking can get pushed over a creative with a higher ROAS (Return on Ad Spend), because Meta wants to keep users on the platform.
We have seen cases where an ad with 2x ROAS gets 80% of the budget while an ad with 4x to 5x ROAS gets starved, because the lower-ROAS ad has better engagement signals. If that bothers you, ABO gives you the override.
ABO vs CBO Testing: Side-by-Side
| Factor | ABO Testing | CBO Testing | |—|—|—| | Budget control | High (you force spend) | Low (Meta decides) | | Testing waste | Higher | Lower | | Learning per dollar | More variables tested | Fewer, but deeper | | Best for | Brands willing to invest in learning | Brands optimizing every dollar | | Audience overlap risk | Higher across adsets | Lower (consolidated) | | Management complexity | Grows fast with scale | Stays manageable | | Typical statistical budget per test | $200 to $300 per adset | Self-allocated by Meta |
The management curve matters more than people admit. As you scale past $5K per day, an ABO structure with 30 to 50 adsets becomes a full-time job. A CBO structure absorbs more spend without proportional management time, which frees you to work on the thing that actually drives results: creative and landing pages.
The 4 Meta Ads Testing Structures That Work in 2026
Here are the four structures we currently run across our client portfolio. Each one matches a specific brand profile.
Structure 1: ABO Testing Campaign + CBO Single-Adset Scaling Campaign
– Testing campaign: ABO with multiple adsets. Each adset = one concept, persona, or angle. Four to six creatives per adset. – Scaling campaign: CBO with one single adset. All winning ads consolidated using post IDs to preserve social proof. – Bidding: Pick one and stick with it. Either highest volume lowest cost throughout, or cost cap and bid cap throughout. Mixing them breaks performance. – Best for: Brands spending $3K+ per day that want maximum control over what gets tested.
This is the closest to the classic playbook. It still works, especially for brands with multiple distinct angles to validate. For the scaling-side detail on bidding, our breakdown of bid cap strategy at scale covers exactly when manual bidding outperforms auto.
Structure 2: One Campaign, One Adset, Test and Scale
– Setup: Single campaign, single adset, all creatives inside. We run 50 to 60 creatives in the same adset for high-spend cases. – Process: Add new creatives, turn off underperformers, scale the campaign budget directly. – Bidding: Highest volume lowest cost is usually fine here. – Best for: Brands spending up to $1K to $2K per day per product. Also works for high-spend single-product brands.
This is the structure that surprised us most post-Andromeda. We expected mixing personas and angles inside one adset to confuse the algorithm. It does not. Meta’s consolidated learning actually performs better with mixed concepts in the same adset, because every additional creative gives the system more data to refine distribution.
The one risk: turning off ads with low ROAS that are actually driving incrementality. Read your data carefully before killing creatives.
Structure 3: One Campaign, Multiple Adsets CBO
– Setup: Single CBO campaign. Each adset = one creative batch, one angle, or one concept. – Process: Meta distributes budget across adsets based on forecasted performance. Scale by raising the campaign budget. – Best for: Brands with clearly distinct angles that benefit from adset-level segmentation, but that still want CBO efficiency.
This is the middle ground between Structure 1 and Structure 2. You get CBO’s automatic budget allocation with slightly more structural control over what learning gets consolidated where.
Structure 4: One Campaign Multi-Adset Testing + One Campaign Single-Adset Scaling
– Testing campaign: CBO with multiple adsets, one per creative batch. – Scaling campaign: Single campaign, single adset, all winning ads via post ID. – Scaling bidding: This is where cost cap or bid cap shines, because you want tight control on the proven creatives at scale. – Best for: Brands spending heavily that want CBO testing efficiency plus a controlled scaling lane.
This is the structure we deploy most often for clients spending $10K+ per day. It separates the high-volatility testing function from the high-control scaling function without creating excessive audience overlap, because the scaling adset uses post IDs that already have engagement history.
For brands struggling to push past plateaus with this kind of structure, our analysis of scaling Facebook ads with frequency and top-funnel covers the volume-side mechanics.
The Engagement Rate Ranking Trick Most Media Buyers Ignore
Inside Meta Ads Manager, custom columns include a metric called engagement rate ranking. Almost no one uses it. They should.
This metric tells you whether Meta considers your ad’s engagement to be below average, average, or above average compared to similar ads competing for the same audience. Ads with average or above-average engagement ranking get distributed to more people and tend to have lower CPMs, because Meta is optimizing for user experience, not just advertiser ROAS.
When we audit underperforming accounts, low engagement rate ranking is one of the first signals we check. According to Meta’s official ad relevance diagnostics documentation, ads with below-average rankings can see materially higher costs per result, which compounds at scale.
If an ad is converting well but has below-average engagement ranking, you are paying a CPM penalty that no amount of bid optimization will fix. The lever is the creative itself.
When CBO Lies to You (and What to Do)
CBO is right most of the time. But not always.
We recently ran a CBO campaign with one adset and a strong winning creative pool. We added more creatives. Performance dropped. We duplicated the campaign, kept the original running, and in the duplicate we kept only the top 10 creatives in a single adset. Performance recovered immediately.
The lesson: CBO’s forecasting can get biased by the new creatives you add, even if the old winners are still strong. When that happens:
- Duplicate the campaign with only the proven top creatives.
- Keep the original live while the new one ramps.
- Compare 7-day windows before deciding which to kill.
This is also why we sometimes run two parallel ad accounts for the same brand at very high spend levels: one optimized for highest volume lowest cost, one optimized purely for cost cap. It removes the bidding-mode interference that breaks single-account hybrid setups.
How to Pick Your Structure: Quick Decision Path
– Spending under $1K per day: Structure 2 (one campaign, one adset, test and scale). – Spending $1K to $5K per day, single product: Structure 2 or Structure 3. – Spending $1K to $5K per day, multiple products or angles: Structure 3. – Spending $5K to $10K per day: Structure 1 or Structure 4. – Spending $10K+ per day: Structure 4, often with split bidding accounts.
Do not blindly copy a structure that worked for another brand. The agencies and brand owners who claim one universal setup beats all others are usually generalizing from two or three accounts. Across 70+ brands, we have seen every structure win in the right context and lose in the wrong one.
FAQ
What is the best Meta ads testing structure post-Andromeda?
There is no single best structure. For brands under $2K per day, one campaign with one adset that handles both testing and scaling tends to outperform. For brands at $5K+ per day, a CBO testing campaign paired with a single-adset CBO scaling campaign using post IDs is the most common winner.
Should I use ABO or CBO for testing in 2026?
Use ABO if you want to force Meta to spend on every angle and persona you want to learn from, and you can absorb the testing waste. Use CBO if you want Meta’s AI to forecast winners and allocate budget efficiently, accepting that some creatives will not get tested.
How many creatives should I put in one adset?
Four to six creatives per adset for ABO testing. For a single-adset test-and-scale CBO campaign, we run 50 to 60 creatives if the budget supports it ($1K to $2K per day minimum at that creative volume).
Can I mix highest volume lowest cost with cost cap in the same campaign?
No. Mixing bidding strategies inside the same campaign or scaling structure consistently breaks performance in our testing. Pick one bidding mode per campaign and stick with it. If you need both, run them in separate campaigns or separate ad accounts.
Why did my CBO stop spending on a winning ad?
CBO uses engagement signals alongside performance, so a creative with high ROAS but lower engagement rate ranking can get starved in favor of a more engaging but lower-ROAS ad. The fix is usually to duplicate the campaign with only the proven winners and let it ramp from a clean slate.
Is the old ABO-to-CBO scaling playbook dead?
No, it still works, especially at higher spend levels and when you have distinct angles to validate. It is just no longer the default winner. Post-Andromeda consolidation has made single-campaign test-and-scale structures competitive or better for many brand profiles.
About the Author
Antonio Ventre is the founder of Skaleit, a Meta ads agency for ecommerce brands. Skaleit manages multi-million-dollar monthly ad spend across 70+ DTC brands and specializes in post-Andromeda testing and scaling structures.
Want Skaleit to Build This System for Your Brand?
If you want a testing and scaling structure built around your specific budget, SKU mix, and growth stage, book a strategy call with our team. We will audit your current setup and map the structure that fits your brand.