Popunder and push are volume formats. You buy impressions (popunder) or clicks (push), pour them onto a landing page, and hope enough convert to clear your costs. The pricing model reflects that: almost everything in this space is sold on CPM or CPC: you pay for the media, and the risk of whether it turns into revenue is entirely yours.
There’s another way to buy it, and in 2026 it’s still rare on pop and push: CPA, pay per conversion. A look at how it works, why so few networks offer it here, and where it’s the right call versus where CPM still wins.
Three ways to pay, and who carries the risk
| Model | You pay for | Who carries conversion risk |
|---|---|---|
| CPM | 1,000 impressions | You, entirely |
| CPC | Each click | Mostly you |
| CPA | Each conversion | The network |
On CPM, a $0.30 popunder buy costs the same whether it produces zero deposits or fifty. On CPA, you only pay when a conversion fires; the network eats the impressions that don’t land. That is the whole appeal, and also why it’s uncommon on broad pop/push: guaranteeing conversions on high-volume traffic is a real bet for the network to make.
Why CPA is rare on popunder and push
Networks avoid CPA on pop/push for defensible reasons:
- Conversion risk sits with them. If the traffic doesn’t convert, they’ve served the media for free. They’ll only offer it where they trust the source quality.
- It needs a clean, trackable event. CPA is meaningless without a reliable postback. A vague “engagement” goal can’t be priced per action.
- It doesn’t fit every offer. Broad, low-intent pop traffic against a long, high-friction funnel is a bad CPA candidate. A tight funnel with a well-defined event (deposit, install, first lead) is a good one.
So when a network does offer CPA on pop or push, it usually comes with conditions: selected inventory, a required conversion signal, specific verticals and geos. Those conditions are what make the model sustainable.
Auto CPA, the middle ground
There’s a second mechanism that gets overlooked. You don’t always want to hand the network a fixed CPA price; sometimes you want to keep buying on impressions but stop hand-tuning bids. That is Auto CPA (target-CPA bidding): you set the cost-per-conversion you’re aiming for, and the system raises or lowers bids automatically to hit it, using the conversions it observes.
The difference in one line:
- CPA pricing: you pay per conversion; the network carries the risk.
- Auto CPA bidding: you still buy the media, but the bidder optimizes toward your target CPA instead of you guessing bids by hand.
Auto CPA works on far more inventory than fixed CPA, because it doesn’t require the network to guarantee anything; it just needs a conversion signal to optimize against.
How Adexium prices pop and push
Adexium runs all three models depending on the format:
- Popunder: from $0.30 CPM, with CPA available on selected inventory.
- In-Page Push: from $0.001 CPC, with CPA available.
- Web Push: from $0.001 CPC.
- Auto CPA (target-CPA bidding): available across all formats, including pop and push, once conversion tracking is in place.
The practical setup: start on CPM/CPC to gather conversion data, wire up a postback, then switch the campaign to Auto CPA so the bidder chases your target cost-per-conversion, or move to fixed CPA where the inventory and offer qualify. Two-layer anti-fraud runs underneath either way, and source-level stats let you whitelist what converts.
Where CPA wins and where it doesn’t
Good fit: a defined conversion event (deposit, install, verified lead), a funnel short enough that pop/push intent can carry it, and a vertical/geo where the network already sees quality. iGaming deposits, app installs, and sweepstakes leads are classic CPA-on-pop candidates.
Poor fit: long multi-step funnels, soft “engagement” goals with no hard event, or brand-new offers with no conversion history for the model to learn from. Here, CPM plus your own optimization still beats a CPA the network won’t quote, or an Auto CPA target it can’t yet hit.
The honest rule: CPA and Auto CPA move risk and manual work off your plate, but they need a real conversion signal and enough volume to optimize. Feed them that, and pop/push stops being a pure guess. Starve them of it, and no pricing model saves a weak offer.
FAQ
Can you really buy popunder traffic on a pay-per-conversion basis? Yes, on networks that offer it. It’s uncommon because the network carries the conversion risk. Adexium offers CPA on Popunder and In-Page Push on selected inventory, plus Auto CPA (target-CPA bidding) across all formats.
What’s the difference between CPA and Auto CPA? With CPA you pay a fixed price per conversion and the network absorbs non-converting impressions. With Auto CPA you still buy the media (CPM/CPC), but the bidder automatically adjusts bids toward your target cost-per-conversion. Auto CPA works on more inventory because it doesn’t require the network to guarantee results.
What do I need before I can run CPA or Auto CPA? A trackable conversion event and a working postback or pixel. The model optimizes toward conversions it can observe, so without tracking there’s nothing to price or optimize against.
Which verticals work best for CPA on pop/push? Offers with a clear, relatively fast conversion (iGaming deposits, app installs, sweepstakes and finance leads) outperform long, high-friction funnels, which are better bought on CPM with your own optimization.
Is CPA traffic lower quality? Not inherently. On Adexium the same two-layer anti-fraud applies, and CPA is offered on inventory the network already trusts. What changes is who carries the risk of a non-converting impression, not the filtering underneath it.
