Last updated: 26 August 2026

What actually happens in the tab when popunder ads fire

A visitor clicks something on a publisher page, a second window loads underneath the one they were reading, and nothing appears to happen at all. That delay is the defining property. Popunder ads reach the visitor minutes later, when the original tab closes and an unfamiliar page sits in front of them with no context whatever. Everything about buying the format follows from that single fact, including why the landing page carries the whole burden and why creative design plays almost no part in the outcome.

How a browser handles popunder ads

Opening a window requires a user gesture, which is why popunder ads attach themselves to a click that was going somewhere else anyway. Browsers enforce this strictly now. Any network claiming delivery without a gesture is describing something older, and pressing on that in a sales call reveals how current the inventory really is.

Built-in blocking removed much of the inventory that existed a decade ago. Extensions removed more. What survives sits mostly on properties where visitors knowingly accept a rougher experience in return for free access, which is the same bargain that underpins most adult traffic supply. Delivery rates vary enormously by market and by device, and again by the individual property inside a given market. The only reliable figure is the one your own tracker records rather than the one a rate card advertises.

Tab focus and what the visitor sees first

Desktop behaviour matches the classic description, with the new window sitting behind and found later. Mobile browsers frequently bring the tab straight to the front instead, turning the format into something far closer to an interstitial that nobody asked for.

That difference changes the page you need. A foreground arrival tolerates a slower reveal, while a background arrival has forgotten the click entirely and needs a plain answer inside the opening line of the page. The network notes on Adult Web Traffic record which operators split their reporting by device at all, and the split matters more than any single placement decision taken afterwards on either device.

Frequency capping that keeps popunder ads viable

One impression per visitor per day is the working default here, and any network permitting more than that is selling irritation rather than reach. The second view of the same page produces almost no conversions and a measurable share of complaints to the publisher, which eventually costs you the placement itself rather than merely the impression. Publishers watch complaint volume closely, because it affects their own relationships downstream. A buyer generating an unusual amount of it gets quietly deprioritised on popunder ads long before anybody sends a warning.

Capping mechanics deserve reading carefully, because a cap applied per placement lets a single visitor meet your page across a dozen separate properties in one afternoon, while a cap applied across the whole account will prevent exactly that from happening. Most advertising platforms document that difference badly, and some never state it.

Pricing and bidding on a thin conversion signal

Almost everything trades on cost per mille, at prices low enough that arithmetic rather than instinct drives the decisions. Tier three clears in cents while tier one runs to several dollars, and the cheap end tempts buyers into markets where the offer has no working payment method available at all. Volume is never the constraint here. Available conversions are the ceiling, and no quantity of extra impressions changes that arithmetic once the offer stops matching the audience underneath it.

Dynamic bidding raises your effective price on inventory the platform expects to perform, which works once enough conversion data exists and behaves randomly before that point. Start flat, gather a full week of source-level results, and only then switch across to it.

Cookie-based capping degrades every year as storage restrictions tighten, which means a stated cap increasingly describes an intention rather than a guarantee, and the difference between those two lands on the invoice. Server-side capping holds up better, and rather fewer networks offer it. Compare that against push ads, where the subscription itself carries the identifier and nothing at all depends on browser storage surviving a clean-up. Ask which method a network uses before the first deposit clears, since answers arrive quickly from operators who built their own stack.

Verticals where popunder ads still convert

Popunder ads suit any offer a stranger understands in a single sentence and acts on immediately. Utilities, streaming access, gaming registrations and sweepstakes dominate the inventory on every network for that reason alone, and the pattern repeats across every geo worth buying at any price.

Sweepstakes and utility offers survive the format because the decision behind them is small enough to take while mildly annoyed, which happens to be the actual emotional state of somebody discovering a window they never opened themselves. Subscription products work when signing up means one tap of confirmation instead of a form with six fields to fill. Anything demanding research, comparison or a conversation with a colleague belongs on a channel where the visitor arrived on purpose and expects to spend some time there.

Considered purchases fail here reliably, because nobody arriving without any context at all evaluates business software, compares insurance quotes or books professional services from a tab they never opened. Native ads carry those offers instead, which is the cleaner division of labour between the two formats.

Landing pages built for a cold arrival

The page is the creative behind popunder ads, since no banner exists anywhere to carry a message, which means the opening line has to say what this is and why the visitor should continue reading before anything else on the screen has finished loading at all. Everything else is negotiable. Audio that plays automatically ends the session immediately, as does any element moving before the text has been read, and both are easy enough to remove long before launch. Load the page on a throttled connection first, then load it again after every design change that follows it.

I worked through the source-blocking examples on popunder-ads.io while rebuilding a client landing template, and what stayed with me afterwards was how far the scroll depth on this particular format sits below every other channel I have ever measured it against. The page now fits a single screen, with the action above the fold and the explanation sitting beneath it for the small minority who scroll down at all.

Splitting by device costs one extra campaign and settles an argument that no blended report will ever manage to settle on its own terms alone. The two environments need different pages, produce different bounce patterns and carry genuinely different honest prices.

The optimisation cycle around popunder ads

Source blacklisting remains the main lever available on popunder ads, and it needs applying weekly rather than daily to work at all. Two hundred impressions tell you nothing at all, while five thousand with no conversions tell you everything worth knowing about that source and its publisher. Weekly cadence also guards against the commonest mistake made on this format, which is cutting a source during one bad afternoon and never learning that it recovered entirely by Thursday on its own.

Keep a record of what was blocked and when, because networks rotate identifiers constantly and a source removed in March reappears in June under a completely different number. The same discipline applies wherever you buy website traffic, though identifiers rotate faster in this corner of the market.

Reading a source report after the first week

Sort by spend, never by conversion rate. The sources costing the most while returning the least are exactly where the money actually goes, and a rate column sorted on its own will hide every single one of them behind tiny placements that happened to convert precisely once. Operating system and browser splits reveal patterns that aggregate reports hide from view completely. Older systems convert poorly on payment offers and acceptably on installs, which justifies separate campaigns rather than bid adjustments inside one.

Cut the worst decile of sources, raise bids on the best performers, and leave the middle alone for another full week. Impatience with the middle is what turns a working campaign into a small one. Most sources need three weeks before their position becomes clear, and campaigns that end up profitable have usually been rebuilt twice by then.