What nobody tells you about MVAS traffic quality in SE Asia — lessons from 8 months in

[Kike]

Hey AF,

Eight months into running MVAS across TH/MY/ID/PL/TR. Made a lot of mistakes, learned a lot of lessons. Sharing this because I wish someone had told me this stuff earlier.

The thing that took me longest to understand:

MVAS isn’t just “mobile CPA.” It’s a completely different animal. The traffic that works for gaming offers, dating offers, or finance offers will often completely fail on MVAS. And vice versa.

The reason is the conversion mechanic. MVAS relies on carrier billing — the user has to be on mobile data (not WiFi), on a supported carrier, and willing to authorize a charge to their phone bill. That’s a very specific user in a very specific context.

What this means for traffic sourcing:

Most traffic sources that claim to “work for mobile” don’t actually work for MVAS. I’ve burned money on:

• Pop traffic (terrible approval rates)
• Social redirect traffic (high click volume, near-zero carrier billing conversions)
• “Premium mobile” DSP packages that turned out to be mostly WiFi users

What actually works:

• Offerwall placements in gaming apps — user is already in a reward-seeking mindset, on mobile data, engaged
• Direct in-app placements in utility apps — lower volume but consistent quality
• Carrier-targeted DSP — only if the platform supports operator-level targeting, not just GEO

The approval rate problem:

This is the metric most people ignore until it bites them. You can have a 10% CR and still lose money if your approval rate is 65%.

Approval rate drops happen when:

• Traffic is coming from WiFi users (carrier billing fails silently)
• Source has incentivized clicks that don’t lead to genuine intent
• Carrier mismatch (targeting wrong operator for the offer)

I now treat anything below 82% approval rate as a hard stop signal regardless of CR.

GEO-specific observations:

Thailand: The gap between AIS and DTAC performance is real and significant. If your DSP doesn’t let you target at carrier level, you’re averaging out two very different audiences.

Indonesia: Volume is massive but source quality variance is the highest of any GEO I’ve run. The same DSP can deliver 12% CR one week and 3% the next depending on which sub-sources are active.

Poland: Slower to convert but higher LTV. Users who do subscribe tend to stay subscribed longer. Worth the lower CR if you’re on a RevShare model.

Turkey: Turkcell users are significantly more valuable than other carriers. Worth paying a premium for carrier-targeted inventory if you can get it.

The question I keep coming back to:

How do you find in-app sources that maintain quality at scale? Every time I find something that works, it degrades within 3-4 weeks. Is this just the nature of the space or am I missing something about source management?

Would genuinely love to hear from people who’ve cracked this.

 

[Pandora92]

Honestly, this hits home. I've been running MVAS campaigns in Indonesia and Vietnam for about eight months now, and yeah—the traffic quality talk is way too polished everywhere.

The big one nobody mentions? Click-to-bill conversion looks amazing on dashboards, but retention is brutal. Like, you'll see 40% conversion rates and think you're killing it, then check back a week later and 80% of those users have already churned. They clicked once, got charged, and either didn't understand the recurring model or just wanted one-off content.

Another thing—carrier filtering is way less reliable than salespeople claim. You think you're getting "premium intent traffic" but so much of it is just users getting pushed through pop-ups or dark patterns they didn't mean to engage. Leads convert but never actually use the service. Makes your LTV look great on paper until month two when everything collapses.

The only thing that's worked for me is being super aggressive with onboarding messages and making the value obvious in the first 48 hours. Also, testing smaller local telcos instead of just the big names—their traffic is often lower volume but way stickier.

Not trying to scare anyone off, just wish someone had told me this before I burned budget learning the hard way. MVAS can work, but you've got to watch retention like a hawk and ignore the shiny first-day numbers.