Notification buying explained : how push ads inventory is built, priced and burned out

Last updated: 31 July 2026

Behind every notification campaign sits a list of people who once pressed allow on a publisher prompt, often years ago and usually without thinking about it for more than a second. That list is the product being sold, and its age decides almost everything that happens downstream of the bid. Push ads stay cheap because the audience cost the publisher nothing at all, and they decay because those same people now receive dozens of messages every single day from every advertiser renting the very same base.

Where push ads inventory comes from and who is actually on the list

Publishers collect the subscriptions that push ads run on through a browser prompt, then sell access to the resulting base through a network of the kind listed on Adult Traffic, which never met either party involved. Nobody sitting on that list is waiting for a message from anybody.

That single fact separates push ads from an email list in every way that matters commercially, and it explains why timing carries far more weight here than relevance ever will in practice. Nobody unsubscribes in the way marketers imagine either, since people simply ignore notifications until a browser reset or a new phone quietly removes them from the base without any deliberate decision at all, which is exactly why reported base sizes and delivered volumes diverge as widely as they do.

Networks report base sizes far larger than the number of people who will ever see a message from anybody at all. A base of ten million may deliver to two million on a good day, and no panel anywhere separates those two figures for the buyer who happens to be paying against the larger of the two numbers displayed on the dashboard.

Collection method shapes quality more than raw volume ever does, and it shapes it in ways that survive every later optimisation anybody attempts. Prompts triggered by genuine interest, much like the pages behind popunder ads, produce subscribers who still respond months afterwards, while prompts disguised as verification checks or download gates build bases that click reflexively and convert at close to nothing at all. The difference surfaces in the second week rather than the first, which is exactly why a three day test against an unfamiliar base proves very little.

Subscription age : the variable that decides what push ads really cost

Freshness dominates every other targeting parameter that push ads offer, and most panels now expose the cohorts directly inside the interface where anybody can filter on them. A subscriber acquired last week behaves nothing like one acquired eighteen months ago. The gap is enormous.

Buying at a flat rate across all cohorts averages two completely different products together and hides both. Fresh delivery costs several times more per click and earns it, while aged delivery looks like a bargain to anyone about to buy adult traffic until the conversion column shows what arrived. I compared cohort performance in my own account against the subscriber age breakdowns published on push-ads.io while restructuring bids, and the collapse point sat within a few days of those figures.

A campaign left on default settings drifts toward older inventory all by itself, since fresh subscribers are always scarce while aged ones stay abundant and very cheap indeed. Nobody touched a setting. Performance decays anyway, quietly, across every zone in the account.

That decay is the commonest complaint about the format and also the most avoidable of them all, because an age ceiling set on the first day removes it permanently at the cost of a smaller audience and a higher average bid on everything that still remains available to buy. Age filtering was a premium feature only three years ago and appears in every serious panel today, so declining to use it is a decision rather than an oversight, and that decision shows up in the acquisition cost inside a fortnight of steady running against a live base.

Delivery timing against the subscriber timezone

Notifications queue while a device sits offline and arrive whenever it reconnects, sometimes several hours later than anybody intended them to arrive. Schedule delivery by the subscriber’s local time rather than by the account’s own clock, or messages will land at three in the morning and bill exactly as much as the ones that landed at seven in the evening to a fully awake audience.

Classic notifications against in page delivery of push ads

Two mechanisms sell push ads under the same name while behaving differently in almost every respect that actually matters to a buyer of them. Classic delivery uses the browser’s own notification system and needs a subscription that already exists, while in page delivery renders a similar looking element inside a website with no subscription involved at all. Only the first faces platform restrictions, and that single fact decides more campaigns than any targeting setting the advertising platforms expose to an advertiser.

Classic delivery disappeared from parts of the mobile market, most visibly on Apple devices, while in page units continue serving there without any difficulty at all. A subscription buy aimed at an audience that cannot hold subscriptions spends its budget somewhere nobody intended.

Performance differs in a direction that follows from the mechanism itself, and the gap is wide enough to plan an entire account around it. In page delivery produces higher click rates alongside weaker intent, since the element appears while somebody is already reading rather than arriving as a separate event they chose to open, and cost per click separates the two approaches as clearly as their behaviour does in the reports.

Which mechanism suits which offer

Classic delivery suits offers that gain from arriving during idle moments, when a phone gets picked up for no particular reason whatsoever. In page suits volume driven propositions where reach matters more than the quality of any single click that arrives. Test both.

Creative construction and rotation schedules that push ads demand

The unit push ads occupy is small, holding an icon, a title, a short body line and a main image on those platforms that support one at all. Every element competes against system notifications a user actually wants, which is not a problem adult web traffic buyers ever face, so plain recognisability beats persuasion inside forty visible characters, and brand names outperform clever phrasing for exactly that reason in almost every test worth running against a live base of real subscribers.

Fatigue arrives faster here than in any other format, and it arrives on a schedule regular enough to plan around rather than react to after the fact. A healthy click rate loses roughly a third of its value within four days of first delivery, whatever sits behind it in the funnel.

Those subscribers meet comparable messages from every advertiser renting the same base, so the decline says very little about the creative itself and a great deal about the noise surrounding it on the device. Rotation is maintenance rather than optimisation, and three variants cycling continuously outperform one strong creative left running until it dies quietly in a report that nobody bothers to open until the end of the quarter.

Where the temptation leads

Deceptive framing is the standing temptation in this format, and it is the reliable route to a closed account. Fake alerts and imitation system warnings lift click rate immediately, then produce complaint rates that get creatives rejected and, on repetition, whole accounts terminated.

Measuring quality when push ads deliver clicks that mean nothing

Accidental clicks are structural to push ads rather than incidental to them, and no setting in any panel removes them from the delivery that an advertiser pays for every day. Notifications appear over whatever somebody happened to be doing at that moment, and a share of every campaign consists of people who pressed the wrong thing and left within a second, which explains most of the price and much of the frustration behind it.

Networks know this and they price accordingly, so the low rate describes quality rather than a bargain, exactly as it does wherever people buy porn traffic. Daily limits belong in the same conversation. Two commercial notifications a day from one advertiser stop being read at all, and the third damages the second.

Bounce rate alone will not separate accidental sessions from genuine ones, because both categories leave a page quickly whenever it fails to load properly on a phone. The ratio between landing page loads and billed clicks works far better as a filter than anything else available in the panel, and anything under seventy per cent points at a source worth pausing rather than tuning any further this quarter or the one after.

Second event tracking settles the question for good, since whatever happens after the first conversion ranks sources in an order that frequently reverses the ranking by click rate entirely. Advertisers buying push ads on click volume alone fund exactly the sources producing the least value of them all.