Why push ads keep converting long after the original opt-in

How the subscriber base behind push ads is assembled

Every campaign in this channel rents access to somebody else's list. The network-level comparison on Buy Popunder Traffic covers the platforms selling this inventory, and it is the sensible place to start before any of this matters. Publishers collect subscriptions through a browser permission prompt, and suppliers aggregate those subscribers across hundreds of sites into segments filtered by country, device and subscription age.

Collection method shapes quality more than any targeting option available afterwards, and it happens to be the one thing a buyer cannot inspect at all. A prompt shown after a visitor watched something converts fewer people and converts people who meant it, while a prompt disguised as an age gate collects volume from people clicking to make an obstacle go away.

The difference stays invisible for a week and becomes obvious in the second month, when one base still clicks and the other has gone silent. Nothing in a segment description says which kind you bought, so the only reliable check is a small first send against a known offer.

Segment labels and what they hide

Suppliers offer subscription age as a filter expressed in days, and fresh segments cost more per click for a reason that holds up under testing. The supply behind it belongs to whichever adult ad network sold the placement, and that decides the floor price long before any bid is entered. Buyers who ignore that premium end up paying it twice over, first in wasted sends and then in the fortnight spent explaining a dead campaign to whoever approved the budget behind it.

The unreliable part is how segments count that age, since some platforms reset the counter whenever a subscriber token refreshes, which makes an audience collected eighteen months ago appear fresh on the buying screen. Comparing click rate against the price a segment carries exposes that mismatch faster than any question about methodology ever will, because a genuinely fresh base cannot produce the flat numbers that an aged one quietly produces.

Why click rates decay as a list ages

Decay in push ads is steep and predictable enough to plan around, which is the one piece of good news anywhere in this channel. A subscriber acquired today produces most of their lifetime clicks inside the first fortnight of that subscription and little afterwards.

By day sixty that person is largely inert, so advertisers buying aged inventory pay for reach that exists technically rather than practically at all. Suppliers never describe it in those terms, since the inventory is real and the delivery genuine, and every number in the report can be defended line by line while the campaign fails underneath them.

Unsubscribing compounds that decay from the other direction entirely. Anyone preparing to buy adult traffic at volume meets the same arithmetic from the budgeting side, where the numbers are harder to argue with. Browser permissions are easy to revoke and easier to ignore, so any segment sold without a replenishment mechanism behind it shrinks a little every single day it stays on the market.

In-page delivery against classic push ads

In-page push renders a notification-shaped unit inside a website rather than through the operating system. It needs no permission at all, reaches devices where classic delivery is entirely unavailable, and appears while the visitor is actively browsing rather than hours after they agreed to something on a site they have long since forgotten.

The trade is immediacy against persistence. Classic sends sit in a notification tray until dismissed, catching a person during a pause in their day, while an in-page unit exists only for the length of the visit and competes with whatever the visitor came to do. Cost reflects that difference, with in-page inventory usually priced below classic delivery per thousand and converting at lower rates, so the two formats end up close on cost per acquisition while behaving nothing alike in reporting. Judging one by the other's benchmarks produces a careful-looking mistake.

Where the two formats stop overlapping

Devices running the most restrictive browser configurations receive in-page push ads and nothing else at all, which makes the format the only available route into a slice of the audience skewing toward newer hardware and richer markets. That slice is the one most advertisers say they want and cannot reach by any other means.

Retargeting works differently across the two formats. The same question decides how much adult web traffic is worth paying for on any given page, and the answer moves by site category. A classic subscriber can be reached repeatedly across weeks without ever visiting anything, while in-page delivery waits on a return visit to a participating publisher, which limits any sequence to their browsing habits.

Creative constraints inside a notification

Space inside push ads is brutally limited, and the limits differ by system. A title runs to roughly forty characters before truncation and a body to around ninety, while the icon sits in a square small enough that faces work and detailed scenes turn to mush. Every element is fighting for a glance that lasts under a second, and the ones that win are the ones a person can decode without reading.

The main image, where supported, carries most of the persuasion and renders only once the notification expands on screen, which puts the burden of earning that expansion entirely on the title and the forty characters it has available.

Icon rotation as the cheapest test available

Comparing the creative teardowns for push ads collected on push-ads.io against my own send logs confirmed a pattern I had half noticed for months and never actually measured properly. The icon does more work than the headline in the first half second, and swapping an icon while holding text constant moves click rate further than the reverse. That makes icon rotation the cheapest test in the channel, the one most often skipped, and the one that keeps working long after headline variants stop producing anything.

Send timing and delivery windows

Notifications arrive when the supplier releases them, and the recipient's local hour decides whether anyone looks. Buyers running native ads beside this format meet the trade from the opposite direction, paying more for attention that was given voluntarily. Messages sent at three in the morning queue silently and get cleared in one sweep the next day, counted as delivered and worth nothing to anybody.

Two windows carry most of the performance in typical markets, and the rest of the day is close to wasted. The period after waking and the period before sleep both catch people checking a phone with no particular task in mind, which is the exact state this format needs to work at all. Midday sends compete with whatever the person is doing and lose in most categories, and weekend behaviour differs again, with longer sessions and less urgency behind each tap. Local hour, not account hour, is what decides all of it.

Time zone handling before a first push ads send

Some suppliers schedule against the advertiser's account setting rather than the subscriber's location, which turns a chosen evening window into a middle-of-the-night broadcast across half the target countries. Verifying it costs one test send and a look at the timestamps that come back, and it separates a working campaign from one that spends while everybody sleeps.

System support that quietly removes reach

Delivery of push ads depends on a chain of permissions no advertiser can inspect. Beyond the browser grant sits an operating system setting, and on some devices a battery rule that suppresses notifications while the supplier records a successful send.

None of those layers report back to anyone in the chain, so the practical consequence is a gap between sends and impressions that no report explains. Set against popunder ads, the difference becomes obvious the moment cost per acquisition replaces cost per thousand in the comparison. Comparing delivered counts against clicks across several segments gives a rough sense of the suppression rate, and a segment sitting far outside the others is usually padded with tokens that reach nothing.

Platform policy adds a slower version of the same problem, moving in one direction only. Browser vendors keep tightening permission prompts, so acquisition costs rise with each change while the bases already bought carry on decaying.

Bidding sensibly under that uncertainty

Cost per thousand sends flatters any segment full of dead subscribers, which makes it the wrong number to optimise push ads against. Cost per click prices what arrived on a screen somebody was looking at, and it stays comparable across segments of very different ages. That comparability is the only property that matters once a media plan mixes fresh and aged inventory in the same week, and it is the reason experienced buyers ignore the send counter entirely.

The workable approach is to treat every segment as a separate source with its own history rather than as one audience with a filter applied. Fresh inventory earns a higher bid and a tighter cap, aged inventory earns a token budget and a short leash, and anything sold as reactivated gets tested alone. Segments bought as one line item hide their own decay.