I spent the better part of 2024 fighting Advantage+ on behalf of clients who had carefully built manual campaign structures. I lost. Not because the clients were wrong to want control, but because Meta had redesigned the entire ad platform around their learning algorithm, and manual bidding on audience segments was increasingly funding the control group in Meta's own experiment.
By late 2024 I stopped fighting and started mapping what Advantage+ actually does well versus where human judgment still outperforms the automation. This article is that map, written after running roughly $1.2 million in Meta spend across e-commerce, SaaS, and local services clients through 2025.
What Advantage+ Actually Changed
Advantage+ Shopping Campaigns launched in 2022 and became the default recommendation for most accounts by 2024. The pitch: Meta handles audience targeting, placement selection, creative rotation, and bid strategy. You provide creative assets and a budget. Here is what it automated away:
| What You Used to Control | What Advantage+ Does Now |
|---|---|
| Audience segmentation | Algorithmic allocation across all audience types |
| Placement selection | Dynamic placement based on predicted conversion |
| Manual bid caps | Automated bidding toward your cost-per-result goal |
| Creative rotation | Performance-based creative serving |
| Frequency caps per audience | System-managed, not user-adjustable |
The practical outcome: the manual 2018-era playbook (tight audiences, separated retargeting campaigns, placement-specific creative, manual bids) stopped working. Not because the logic was wrong, but because the system was no longer operating on that logic.
What Stopped Working First
Granular audience segmentation was the first casualty. Separating lookalike audiences from interest stacks from retargeting pools into individual ad sets made sense when Meta's algorithm needed human help with budget allocation. Once Advantage+ was doing the allocation with real-time conversion signal, my hand-built audience fences were restricting the system from finding the cheapest conversions.
Manual bids went second. I had a client in the home fitness niche who was convinced that bid caps were protecting his CPA. What they were actually doing: throttling delivery on his best-performing creative during peak conversion windows. We removed the caps. His CPM went up 8% and his overall CPA dropped 22% because the system could bid aggressively when the signal was strongest.
Retargeting as a separate campaign structure went third. Dedicated retargeting campaigns with customized messaging for cart abandoners, past visitors, and customer lists still feel like they should add value. In most accounts I have audited since 2024, they cannibalize spend from Advantage+ without improving aggregate ROAS. The exception is discussed below.
What Still Works
Creative is the only remaining lever with real impact. This is not a platitude. It is the practical reality of how Advantage+ operates. The algorithm competes for the cheapest conversion-probability impression. You cannot control who sees your ad or when. You can control what the ad says and shows. That makes creative strategy the highest-leverage activity in Meta Ads in 2026.
The accounts beating benchmarks are running the most creative iterations: three to five new concepts per month, each with a distinct hook (the first three seconds of video or the first line of static copy), tested inside a single Advantage+ campaign alongside existing winners. The algorithm promotes the new winner automatically. You retire the concepts that lose.
Catalog structure still matters for e-commerce. Product feed quality, category accuracy, and pricing are all outside Meta's automation. A clean feed with rich product descriptions and accurate inventory signals improves Advantage+ performance measurably because the system serves catalog ads dynamically and cannot compensate for poor inputs.
Conversion event quality drives everything else. Pixel implementation, the Conversions API (CAPI), and event deduplication are all human responsibilities. An account where 80% of conversions are matched server-side through CAPI will outperform an identical account relying on browser-only tracking. CAPI integration partially restores the attribution signal Meta lost after iOS 14.
Exclusion lists are chronically underused. Advantage+ does not exclude your existing customers from cold prospecting by default. A 90-day purchaser exclusion, uploaded as a custom audience and set as a campaign exclusion, measurably reduces wasted spend on people who already converted.
The Campaign Structure I Use in 2026
For most accounts I now run a simplified three-campaign structure:
Account
├── ASC (Advantage+ Shopping): 60-70% of budget
│ ├── Creative set A: proven winners (3-5 assets)
│ ├── Creative set B: new concepts under test (2-3 assets)
│ └── Exclusion: 90-day purchasers
│
├── Retargeting (Manual): 15-20% of budget
│ ├── Audience: video viewers 50%+ and site visitors, 30-day window
│ └── Creative: direct response, offer-focused, short copy
│
└── Brand Awareness (Manual): 10-15% of budget
├── Audience: broad interest, no lookalike
└── Creative: problem-aware, story-driven
The ASC handles acquisition. The retargeting campaign exists not because it beats Advantage+ on audience allocation (it does not), but because offer-based messaging (discount, urgency, testimonial) should not pollute the ASC creative pool with bottom-funnel content. The brand campaign is almost always a client request rather than a performance driver.
The Numbers Behind the Recommendation
Across eight accounts migrated to this structure in 2025:
| Metric | Before Migration | After 90 Days |
|---|---|---|
| Average CPA | $38.20 | $29.40 |
| ROAS (purchase value) | 2.8x | 3.6x |
| Creative concepts tested per month | 1.2 | 4.1 |
| Active campaigns per account | 9.4 | 3.0 |
The CPA improvement is real but not entirely from Advantage+. Simplifying the structure forced me to concentrate creative effort rather than spreading it thin across nine campaigns. That shift in creative velocity drove most of the improvement.
What I Got Wrong
I underestimated how much Advantage+ would improve for non-e-commerce accounts. My early tests were almost entirely in retail and direct-to-consumer, where the purchase signal is clean and the catalog feed gives the algorithm something concrete to optimize against. I assumed lead gen would be harder, and for the first six months it was.
By mid-2025, ASC for lead gen (using the "leads" objective) had improved enough that several of my service-business clients saw 30-40% CPL reductions by moving to it. The key condition: you need at least 50 conversions per month for the algorithm to have adequate signal. Below that threshold, Advantage+ is essentially guessing.
I also underestimated the importance of CAPI early on. The accounts where I prioritized server-side matching through CAPI saw measurably better learning phase performance than those running on pixel alone. Retrofit this to existing accounts even when everything appears to be working.
When NOT to Use Advantage+
Three situations where I still run manual campaigns:
New accounts with no pixel history. The algorithm needs conversion signal to optimize. Without historical data, Advantage+ will spend the first $2,000 to $5,000 collecting signal rather than converting. For new accounts, I run manual campaigns with tight targeting to seed pixel data before handing control to the automation.
Low-volume, high-ticket B2B offers. If a client closes four deals a month at $25,000 each, the conversion signal is too sparse for algorithmic optimization. Manual targeting and human judgment on creative fatigue works better at this volume.
Brand-sensitive regulated categories. Health, finance, and some regulated categories have placement restrictions that Advantage+ sometimes mishandles. Manual campaigns give you more control over ad placement when category-level brand safety is a genuine concern.
The Boring Reality
The most valuable work in Meta Ads in 2026 is creative production and data hygiene, not campaign architecture. The full-service agency model that charged for audience strategy, bid management, and campaign structure is largely automated. What remains is the judgment work: deciding what creative hypothesis to test next, reading the data to know when a concept is fatiguing, maintaining product feed quality, and ensuring CAPI is firing correctly.
If you are running Meta Ads without a monthly process for producing and testing new creative concepts, you will plateau regardless of campaign structure. The algorithm is excellent at optimization. It cannot generate a new hook for you. That is still your job.