Four million advertisers. Quadrupled in six months. And $60 billion — not Meta’s revenue, but the annual value of ad campaigns now fully automated by AI.
The 22% ROAS Lift Isn’t a Guarantee
Meta markets Advantage+ Shopping Campaigns with a headline claim: 22% higher ROAS versus manually managed campaigns. The number is real. The context matters more.
Independent analysis from VXTX Performance Marketing shows 22% is a top-quartile outcome. The advertisers hitting that number share specific traits: rich conversion history in their pixel, high-quality first-party audience lists, and a creative library large enough for the model to run meaningful tests. For everyone else, median results sit much closer to parity with well-optimized manual campaigns.
Meta’s own data is more instructive when read by layer:
- GEM (Generative Ads Recommendation Model): 3.5% lift in Facebook ad clicks in Q4 2025
- Instagram Reels conversion model: 3% conversion rate improvement
- Incremental attribution model: 24% increase in incremental conversions versus standard attribution (Meta Newsroom, 2026)
Not small gains. But these are infrastructure improvements, not guarantees applied uniformly across every account.
When AI Does the Work, What’s Left?
The Advantage+ model is straightforwardly unsettling: paste a product URL, set a budget, and the system generates creative assets, selects audiences, allocates placement, and adjusts bids — simultaneously across Facebook, Instagram, Messenger, and WhatsApp.
This isn’t a roadmap. It’s Q1 2026.
eMarketer analyst Jeremy Goldman put it plainly: “Meta has quietly turned AI into margin — ad tools are sharper, targeting smarter.” That’s the optimistic read. The practical read: when automation handles targeting and bidding for everyone equally, what’s left to differentiate you is creative quality and first-party data.
Meta is putting $70-72 billion into AI infrastructure in 2026 (Marketing Dive). Q3 2025 revenue grew 26% year-over-year to $51.24 billion, with AI ad tools as the primary driver. Video generation tools alone hit a $10 billion revenue run rate in Q4 2025, growing three times faster than overall ads revenue (Meta Newsroom, 2026).
The machine keeps improving. The question is not whether AI can perform. The question is what you’re feeding into it.
The Southeast Asia Data Gap
In Vietnam and across Southeast Asia, Facebook and Instagram remain the dominant digital advertising channels for most businesses. Yet Advantage+ adoption is low — not because marketers aren’t aware of it, but because the underlying data infrastructure isn’t ready.
Three specific gaps compound each other:
Conversion volume: Advantage+ needs roughly 50 weekly conversion events to optimize reliably. Most SMBs in the region log 20-30 per month — not enough for the model to find a stable learning curve.
First-party audience data: Without a quality customer list as a Lookalike Audience seed, Advantage+ has no starting signal. The AI doesn’t invent high-intent audiences from scratch.
Creative inventory: The model can’t run meaningful A/B optimization across one or two assets. It needs variation — different formats, angles, hooks — to identify what actually resonates with each segment.
Vietnam’s AI market is forecast to grow from $161 million in 2025 to $1.83 billion by 2034, a 31% CAGR (Nexdigm, 2026). That growth only translates to real business outcomes when there’s quality data for AI to learn from. Without it, automation doesn’t improve results — it accelerates budget burn with less human oversight.
Creative Is the Only Remaining Moat
When AI executes targeting and bidding at equivalent quality for everyone in your category, creative becomes the only differentiator left.
Meta’s video generation tools are growing at three times the pace of its overall ad business. AI can produce video ads at scale. But a brief built on genuine customer insight, a consistent brand voice, and messaging tuned to specific pain points still requires someone who understands the customer’s world.
The era of winning by finding a better audience than your competitor is ending. AI closes that gap on both sides. What follows is an era where you win by saying the right thing — to the right person, at the right moment, with a message your competitor didn’t have the customer knowledge to write.
For marketers globally, and especially in markets like Vietnam where the data foundation is still early-stage, the priority isn’t chasing AI tools. It’s building the inputs those tools need to be worth using.
NateCue's Take
The 22% ROAS lift is real, but it's conditional. Advantage+ learns from your existing data. Feed it thin pixel coverage and two creative assets, and it won't save your campaign — it'll just lose budget faster at scale. The smarter question isn't whether to use Advantage+ but whether your data foundation is ready for it: 50+ weekly conversions, a seeded first-party audience, and enough creative variation to let the model test meaningfully. In Southeast Asia, where most SMBs run 20-30 conversions per month, that gap is the actual strategic problem — not AI access. Fix the foundation. Then let automation compound it.