Meta's $3 Billion Scam Ad Problem
March 2026 | Croupier Blog
In December 2025, Reuters published an investigation based on internal Meta documents revealing that approximately $3 billion — 19% — of Meta's China advertising revenue came from scam ads. Illegal gambling, prohibited products, and fraudulent offers running across Facebook, Instagram, and WhatsApp.
Meta internally labeled China its top "scam exporting nation," responsible for 25% of all scam ads globally. The company created an anti-fraud team that cut problematic ads from 19% to 9% of China revenue in the second half of 2024.
Then they disbanded the team.
What Happened Next
After Zuckerberg's input, Meta lifted a freeze on new Chinese ad agencies and reinstated roughly 4,000 suspended agencies, unlocking approximately $240 million in annualized revenue — about half tied to policy-violating ads.
By mid-2025, banned ads climbed back to 16% of China revenue. Meta was serving as many as 15 billion "high-risk" fraudulent ads per day, generating roughly $7 billion annually from problematic advertising.
Meta required a 95% confidence threshold before banning fraudulent advertisers. Those below the threshold continued operating.
An internal audit of 800 China-based ad accounts uncovered approximately $28 million in monthly violating spend. Three-quarters of that spend flowed through protected partners.
Why This Matters for Advertisers
If you're an advertiser buying impressions on Meta, your ads run in the same ecosystem as 15 billion daily fraud ads. Your brand appears alongside scam offers. Your metrics are inflated by the same system that tolerates $7 billion in fraudulent spending because removing it would miss the quarter.
This is the same dynamic that drives made-for-advertising sites in programmatic. Revenue from junk becomes the baseline. Removing junk means missing targets. Each quarter's fraud becomes the floor for the next.
Meta isn't unique. It's just the case where the internal documents leaked. Every platform that sells self-measured inventory faces the same incentive: tolerate enough fraud to hit revenue targets, enforce enough to avoid headlines.
The Measurement Implication
When the platform selling impressions also decides which ads are legitimate, the conflict of interest is structural. Meta's anti-fraud team was effective — they cut scam ads by more than half. But "effective" was bad for revenue, so the team was dissolved.
Advertisers relying on platform-reported metrics can't see this. Their dashboards show impressions served, clicks generated, conversions attributed. The dashboard doesn't distinguish between a conversion driven by a legitimate impression and one that happened despite running next to 15 billion scam ads.
Advertiser-Side Verification
The alternative is measuring at the conversion endpoint. If the advertiser issues cryptographic coupons through their ad channels and counts what comes back, they measure actual performance independent of platform reporting.
Publisher A (a Meta campaign targeting SaaS buyers) redeems 60 out of 1,000 coupons — 6% conversion rate. Publisher B (a different Meta placement) redeems 2 out of 1,000 — 0.2%. The advertiser sees the difference because they're counting their own signed tokens, not reading Meta's dashboard.
The coupons don't fix Meta's scam problem. But they give the advertiser an independent signal that doesn't depend on the platform marking its own homework.
Croupier is a blind relay for cryptographic coupon books. Learn more or request early access.