Phantom Reach: How a Sports Betting Brand Burned $87K Targeting the Wrong Continent

The Situation

A licensed sports betting operator based in the Philippines had secured regulatory approvals for three Southeast Asian markets: Philippines, Thailand (grey market), and Vietnam (grey market). They allocated $87,000 over eight weeks to Meta Ads with the goal of acquiring 1,500 first-time depositors (FTDs) at a target CPA of $58.

The media buying team configured campaigns using interest-based targeting: users interested in "sports betting," "online gambling," and "live odds." They layered a 2% lookalike audience built from their existing customer database.

What Went Wrong

The lookalike seed contained 12,000 email addresses — but nearly 40% belonged to users who had registered from India, Bangladesh, and Pakistan during a previous global campaign. Meta's algorithm optimized delivery toward these cheaper impressions, since CPMs in South Asia were 60% lower than in the Philippines.

Within two weeks, delivery reports showed 63% of impressions going to India and Bangladesh — markets where the operator held no license and could not legally process deposits. The remaining 37% split unevenly between the Philippines and Thailand.

The team noticed low conversion rates but attributed it to "creative testing phase" rather than audience contamination. They continued spending for another six weeks.

Diagnosis

RedClaw's audit revealed three compounding errors:

1. Poisoned lookalike seed — The customer list was never filtered by jurisdiction. Meta's algorithm found the cheapest path to "similar users," which led to non-regulated markets.
2. No geo-exclusion layers — The campaign used broad Southeast Asia targeting with no negative country exclusions.
3. Wrong optimization event — Campaigns optimized for "Registration" rather than "First Deposit," rewarding bot-heavy markets where registrations were cheap but deposits were impossible.

The Fix

We rebuilt the entire audience architecture:

• Exported a clean customer list filtered to users with at least one verified deposit from licensed jurisdictions
• Created 1% and 3% lookalikes from this sanitized seed, restricted to Philippines only (the only market with clear legal standing)
• Added country-level exclusion lists covering 47 non-target markets
• Shifted the optimization event from Registration to the custom "FTD" event fired after first deposit confirmation
• Implemented a 72-hour creative rotation to prevent the algorithm from finding low-quality delivery paths

Results

Within 21 days of relaunching:

• ROAS climbed from 0.4 to 3.9
• CPA dropped from $112 to $33 per first-time depositor
• CTR improved from 0.6% to 1.9% as ads reached genuinely interested bettors
• The client acquired 847 FTDs in three weeks — more than they had generated in the previous eight weeks combined

The core lesson: in regulated industries, audience architecture is not a performance lever — it is a compliance requirement. Targeting the wrong users does not just waste budget; it generates legal exposure.

Early Warning Signals: How to Catch Geo-Contamination Before It Burns Budget

Audience contamination rarely announces itself. In accounts we've audited, the typical pattern is a campaign that looks "fine" on blended metrics while the delivery report tells a completely different story. Four signals show up again and again:

1. The delivery breakdown drifts away from your license map. Open Ads Manager, break down delivery by country, and compare it against the list of markets you are actually licensed (or tolerated) to operate in. Any meaningful impression share outside that list is spend you cannot convert. Most teams never run this report because campaign-level metrics don't force them to.
2. Registrations look healthy while deposits collapse. Healthy iGaming funnels convert roughly 25–45% of registrations into first deposits. When registration volume grows but reg-to-FTD falls under 15%, the traffic source has usually changed underneath you — often toward markets where users can register but cannot legally or practically deposit.
3. CPMs are suspiciously cheap. A sudden CPM drop feels like a win, but the algorithm buying dramatically cheaper impressions usually means it found a lower-cost geography or placement, not a better audience. Cheap reach into non-regulated markets is worth exactly zero.
4. Lookalike audiences built from raw CRM exports. If the seed list was never filtered by jurisdiction and deposit status, the lookalike inherits every contaminated profile in it. The algorithm then scales the contamination faithfully.

The Recovery Playbook

If you recognize this pattern in your own account, the fix sequence that consistently works looks like this:

1. Quantify the leak first. Pull a 30-day country-level delivery report and calculate the exact spend share going to non-licensed markets. This number decides whether you patch or rebuild.
2. Sanitize the seed. Export your customer list, filter to users with at least one verified deposit from a licensed jurisdiction, and rebuild custom audiences only from that subset.
3. Rebuild one market at a time. Launch lookalikes restricted to a single licensed country before expanding. Multi-country lookalikes give the algorithm room to chase cheap impressions again.
4. Layer explicit country exclusions. Do not rely on inclusion targeting alone — add a negative country list covering every market you cannot serve.
5. Move optimization to the deposit event. Registration-optimized campaigns reward exactly the markets you need to avoid. Expect a fresh learning phase of roughly 50 conversion events after the switch.
6. Verify daily for two weeks. Check the country breakdown every day until delivery stabilizes inside your license map.

Prevention Checklist

• [ ] Country-level delivery breakdown reviewed weekly, not just campaign-level CPA
• [ ] Every seed list filtered by jurisdiction AND verified deposit before upload
• [ ] Negative country exclusion list maintained and applied to all new ad sets
• [ ] Campaigns optimize to the deposit event, never registration
• [ ] Compliance sign-off required before entering any new market
• [ ] Internal alert when more than 5% of impressions land outside licensed markets

Where Your Numbers Should Be

Once delivery is clean, benchmark against the market: median iGaming CPA on Meta runs around $45 per first-time depositor, with top-quartile accounts reaching 8.5x FTD ROAS. If your post-fix numbers sit far below that, the next suspects are creative fatigue and tracking loss rather than targeting. Compare your metrics against the iGaming Meta Ads benchmarks and the full iGaming ROAS Benchmarks 2026 breakdown.

Related reading: Meta Lookalike Audience Guide · Advanced Meta Ads Audience Targeting · Another angle on the same failure type: How Blind Targeting Burned $50K in iGaming Meta Ads