The best Meta ads strategy in 2026 for ecommerce brands is a one-CBO-per-hero-product structure with 20 to 50 ads inside each ad set, weekly creative launches, and no manual audience suggestions. That is the short answer we give every founder who asks us how to structure campaigns after Meta’s Andromeda update. In this Q&A, we answer real questions from ecommerce operators running jewelry, skincare, fashion, and luxury brands, and we show exactly how we apply this framework at Skaleit across hundreds of thousands in monthly ad spend.
Meta ads (paid advertising on Facebook and Instagram through Meta Ads Manager) has changed dramatically with the Andromeda algorithm update, and most of the advice online is outdated. As of 2026, the platform optimizes at the ad set level as a portfolio, not at the single-ad level, which changes how you should think about structure, testing, and creative rotation. Let’s dig into the questions.
TL;DR
– Structure: One CBO (Campaign Budget Optimization) per hero product, maximum 3 hero CBOs, plus one catalog CBO for broad and one for retargeting. – Ad density: 20 to 50 ads per ad set, not 5 or 6. Meta’s post-Andromeda ad structures optimize the whole set, not individual ads. – Optimization: Always optimize for purchases, highest volume. Skip cost cap and target ROAS unless you accept low spend. – Targeting: Fully open, no interests. Exclude existing customers only if spend on them gets excessive. – Testing cadence: Launch new creatives weekly or bi-weekly, never daily. – Luxury brands: Storytelling, community, and celebrity/influencer signals matter more than ad mechanics.
Jewelry and Fashion Brands: One CBO per Hero Product
Jewelry and fashion brands almost always have a Pareto distribution in their catalog. We routinely audit stores where 2 or 3 SKUs generate 80% of revenue while 40+ SKUs share the leftover 20%. The winning Meta ads strategy for these brands ignores the long tail and doubles down on the winners.
Here is the exact structure we deploy:
– CBO 1: Hero Product A, one ad set, 20 to 50 ads inside – CBO 2: Hero Product B, one ad set, 20 to 50 ads inside – CBO 3: Hero Product C, one ad set, 20 to 50 ads inside (only if the SKU justifies it) – CBO 4: Dynamic catalog ad, broad targeting, entire catalog – CBO 5: Dynamic catalog retargeting, entire catalog
For a fashion brand with variants (colors, sizes, prints), you build one CBO around the winning design, then split ad sets only if you want to isolate colorway performance. In most cases, one ad set per CBO with high ad density outperforms multiple ad sets with low density. This mirrors the approach in our fashion drop launch playbook, which sells out drops in 48 hours using the same principle.
Each hero-product CBO becomes a scaling vehicle. The end state is one campaign per hero product spending as much as the audience can absorb profitably.
Skincare Brands With 30+ SKUs: Focus Your Meta Ads Strategy
Skincare brands with sprawling catalogs make the same mistake as fashion brands: they try to advertise everything. The fix is identical. Isolate the 2 to 3 SKUs driving the majority of revenue, build one CBO per hero SKU with high ad density, and let a catalog CBO handle the rest for both broad and retargeting.
We covered the full framework in our skincare Meta ads strategy that hit 5.86x ROAS. The settings never change:
– Objective: Sales, optimized for purchases (never brand awareness, never traffic) – Bid strategy: Highest volume, not cost cap or target ROAS – Placements: Advantage+ placements on – Targeting: Fully open, no interests, no lookalikes
Why highest volume and not target ROAS? Because ecommerce is a volume game. Spending $2,000 a month at 5x ROAS is a hobby. Spending $50,000 a month at 3x ROAS on stable contribution margins is a business. Cost caps and target ROAS caps also starve the algorithm of learning data, which compounds the problem over time.
The one exception: bid cap on your very best performing set of ads. We are currently scaling a client aggressively with a bid cap sitting on top of a large ad set of proven creatives. It is not a replacement for highest volume, it is a scaling layer on top.
Targeting Post-Andromeda: Keep It Fully Open
Broad targeting is the only targeting that works reliably in 2026. This is not opinion, this is how Meta’s Advantage+ system is built to operate. When you add interest suggestions, Meta expands past them anyway, so you are adding a signal that gets ignored or, worse, biases early delivery in the wrong direction.
The only exclusion worth using is existing customers, and only when their share of spend gets unreasonable. Do not exclude website visitors or engagers by default. Under Andromeda, Meta is building one funnel across all awareness stages, and the data from existing buyers feeds cold prospecting through the GEM (Generative Media) delivery system. Exclusions break that data loop.
This logic pairs with our broader work on post-Andromeda targeting methods, where creative diversity replaces audience segmentation.
Ad Set Ad Density: Why 50 Ads Beat 5
Ad density inside each ad set is the single biggest structural shift under Andromeda. Meta is no longer optimizing the best individual ad. It is optimizing the best ad set as a portfolio, considering both ROAS and spend distribution across the ads inside it.
Our rule: pack 20 to 50 ads into each ad set, spanning different user awareness stages and different creative angles. Statics, videos, UGC, testimonials, product demos, problem/solution hooks. The more diverse the creative pool inside one ad set, the more surface area GEM has to serve the right ad to the right user at the right moment.
When the algorithm has to choose between showing a top-of-funnel hook video to a cold user and a testimonial static to a repeat visitor, both need to live in the same ad set for Meta to make that choice fluidly. This is why the Andromeda-native strategy minimizes ad set count and maximizes ads per set.
When to Add New Ads to an Existing Ad Set vs. Launching a New One
– Existing ad set performing well: Do NOT add ads. Launch new ads inside a new ad set to protect the learning. – Existing ad set underperforming: Add new ads inside it. You are already resetting learning by turning off losers, so add fresh creative in the same motion. – Existing ad set decayed with no active winners left: Turn off the entire ad set and launch a new one with a fresh batch.
Testing Cadence: Weekly, Not Daily
New creative launches should happen weekly or bi-weekly, never daily. Daily launches prevent GEM from stabilizing on any signal, and you burn budget on constant re-learning. According to Meta’s own learning phase documentation, ad sets need roughly 50 optimization events to exit the learning phase, and every new ad interferes with that process.
Weekly launch cadence is the sweet spot. It gives the algorithm 7 days of stable delivery, lets you evaluate the batch fairly, and refreshes the pool before ad fatigue hits.
Luxury Brands: Storytelling Beats Ad Mechanics
Luxury brands operate on different economics. A $300 cream and a $50 cream can contain nearly identical formulations. A $500 t-shirt and a $70 t-shirt can come from the same factory. The price gap is paid entirely for identity, community, and story.
This means the Meta ads strategy playbook shifts:
– Community first, ads second: You need an identity people want to belong to before ads will scale. – Celebrities and influencers: Signal that the brand belongs in a specific status tier. This lives outside Ads Manager but changes how ads perform inside it. – Storytelling creative: Ads should sell the story, the founder, the world of the brand, not the product spec. – Emotional outcome: How does the customer feel wearing or using the product? That feeling is the entire pitch.
We expand this framework in our luxury brand Meta ads strategy post-Andromeda. If the brand foundation is not built for luxury positioning, no ad structure will fix it.
Comparison: When to Kill an Ad Set vs. Individual Ads
| Situation | Action | |—|—| | One ad losing money, ad set overall profitable | Kill the single ad | | Multiple ads losing, ad set still profitable | Kill individual losers, add fresh ads inside | | Ad set overall unprofitable, few active ads left | Kill the whole ad set, launch a new one | | Ad set profitable but plateauing | Leave it, launch a new ad set with fresh batch |
The goal is always to protect ad sets that are learning while cycling out those that have decayed past the point of recovery.
FAQ
About the Author
Antonio Ventre is the founder of Skaleit, a Meta ads agency for ecommerce brands. Skaleit has spent over $10M on Meta ads across fashion, jewelry, skincare, supplement, and luxury brands, and specializes in post-Andromeda scaling strategies.
Want Skaleit to Build This System for Your Brand?
If you are running a jewelry, fashion, skincare, or luxury ecommerce brand and want us to build the full CBO structure, creative testing system, and scaling framework for you, book a call with Skaleit. We only take on brands we know we can scale profitably.