Popunder

What it is

Full-page ad opened in a new tab behind the active window. Triggered by a user click on the publisher page, so it bypasses most ad blockers. Runs on every device, every OS.

I do not specialise in popunder. Five years at Mobidea I sat next to the popunder team — I read their dashboards, ran a few parallel-buy tests against push for the same Tier-1 iGaming offers in Q3 2023 (n=2.1M impressions, US + UK + DE). What follows is what the data showed, not the rate-card story.

What the numbers actually say

• eCPMs: Tier-1 popunder cleared at $1.40–$2.80 median across iGaming offers, Q3 2023 (n=2.1M). Push for the same offers cleared at $0.90–$1.80 median. Popunder pays more per impression.
• Day-1 vs day-7 CR: popunder CR on day 1 ran at 0.42–0.58% in the sample. Day 7 stabilised at 0.18–0.31%. Roughly half of the day-1 conversions don’t survive a 7-day attribution window. Push, for comparison, had a tighter day-1-to-day-7 gap (0.34% → 0.27%) in the same test.
• The bell curve is bimodal: top-decile popunder sub-sources delivered 58% of human conversions; the bottom 12% delivered fraud and broken-fingerprint signal. Aggregated CPM averages hide both ends.

When popunder is the right format

• iGaming player acquisition where the LTV math survives a 50–60% day-7 conversion-decay rate
• VPN, antivirus, utility installs — short consideration, single-session conversion
• eCommerce remarketing on warm audiences
• High-volume Tier-2/Tier-3 scaling where push subscriber bases are thin

When it is the wrong format

• Anything with a multi-touch attribution window beyond 7 days
• B2B SaaS, anything with a 30-day sales cycle
• Offers where the day-7 CR halves the day-1 CR — the lift is auction warm-up, not real demand

Specs

• Pricing models: CPM · CPA
• Devices: Mobile + Desktop, every OS
• Geo coverage: 240+ countries
• Anti-fraud: Multi-layer filter, ML scoring per impression. Treat the ML score as a posterior probability, not a verdict — overlay a fingerprint check on your own end.

The marketing deck calls popunder “premium traffic.” The publisher-quality histogram calls it bimodal. Both can be true. Decide on the second week of data, not the first.