Why push ads decay faster than any other display channel

By Smith Jones · 25 August 2026

Push ads deliver a message to a browser that already agreed to receive notifications, and that makes them the only cheap format with a permission trail behind it. Delivery costs a small fraction of one cent in most markets. The problem is arithmetic. A subscriber who joined yesterday clicks roughly ten times more often than one who joined six months ago, and every base ages continuously. Performance therefore depends less on targeting than on how recently the list was collected and how hard it has been worked since.

How a subscriber base behind push ads is assembled

Every name on a subscriber list came from a permission prompt somewhere, and the context of that prompt determines everything downstream. A prompt shown on a content site after somebody started reading produces a different subscriber than one shown behind a fake download button.

Collection methods behind a subscriber list vary from honest to deceptive inside one platform. Some publishers ask plainly, some overlay the prompt on a video play button, and some run pages that do nothing except request permission. The last group builds volume fastest and produces the worst engagement, because nobody clicking allow in that context expected messages afterwards. I read through the collection policy descriptions on push-ads.io while working out why two sources at identical prices behaved nothing alike, and the difference sat entirely in how each base had been gathered.

Browser rules that shaped the format

Chrome and Firefox both moved to quieter permission interfaces once abuse became widespread, replacing the interrupting dialogue with a smaller indicator. Opt in rates fell sharply afterwards. The subscribers who remained were more deliberate, which raised the average quality of a push ads name.

Safari never supported web push on iOS the same way, while support arrived later through a different mechanism, which requires the site to be added to the home screen before anything can be delivered at all. That gap explains why this inventory, like most adult traffic sources, skews toward Android and toward markets where Android share is high, and a campaign priced on a global average finds its delivery concentrated in places nobody targeted deliberately. Checking the device breakdown after the first thousand impressions catches it early. Explicit country targets prevent it entirely.

Classic and in page push ads compared

Two push inventory products share this name and behave differently in almost every respect. Classic push reaches a subscriber through the operating system notification layer, arriving whether or not the person is browsing. The message waits on the lock screen until it is dismissed.

In page push renders inside a webpage as a lookalike element, reaching whoever happens to be on the page at that moment, with no permission involved and no record of one. The distinction matters during a dispute, because a consent record answers a complaint and an absence of one does not. Buyers who mix both formats inside a single campaign lose the ability to answer either question cleanly, and the gap becomes visible at the worst possible moment, when a payment provider asks how the audience was obtained.

Why in page volume looks unlimited

In page push ads inventory is limited only by pageviews, so it never runs out the way a subscriber base does, and advertising platforms lean on it whenever classic supply tightens. Separate the two formats into different campaigns, or one disguises the economics of the other.

Timing and frequency limits for push ads

Click through rates on it look impressive and collapse under examination, because a notification shaped element near the top of a mobile page collects accidental taps at a rate no legitimate creative could earn. Delivery timing separates profitable push campaigns from expensive ones, and the relevant clock belongs to the subscriber rather than to you. Morning and early evening windows carry most of the profitable volume, while a notification sent at three in the morning queues silently and arrives in a stack of eleven others when the device wakes.

Platforms selling push ads differ in whether they respect local time zones by default, and several do not. Check that single setting before launch. Server time and subscriber time diverge by up to twelve hours across a wide target, which is enough to put an entire day of budget in front of sleeping devices.

Frequency caps and the cost of an unsubscribe

Every impression carries a small chance the subscriber turns notifications off for good. Publishers therefore cap delivery, usually between three and eight messages a day. A buyer paying for volume, or anyone who came here to buy adult traffic, competes for those slots against everyone else, which is why bidding higher wins earlier slots, and the first message somebody sees after waking outperforms the fourth by a margin wide enough to justify the premium.

Why push ads click through rates fall over time

Decay in push campaigns follows subscriber age with a regularity that makes it forecastable. A base collected within the last seven days commonly clicks at rates between one and three percent. The same base at thirty days runs closer to half a percent. By ninety days most remaining names have either unsubscribed or learned to dismiss without looking. The curve is steep at the beginning and almost flat afterwards, which is why the difference between a seven day base and a fourteen day base matters far more than the difference between sixty days and ninety.

Platforms segment inventory by age because the difference between one segment and the next decides the campaign. Fresh segments cost several times more per click and earn it back on conversion rate rather than on volume, so buying on click price alone selects for the worst names available on the platform.

Reading a platform's age labels sceptically

Age labels depend on the platform recording a subscription date honestly, and nothing external forces that record to be accurate or keeps it current. A base sold as fresh that produces click rates matching a ninety day segment has been mislabelled, and finding out costs one small campaign run against a reference creative you keep unchanged across every account. That single control turns a vague suspicion into a number.

Fraud and recycled bases in push ads inventory

Fraud here concentrates in the subscriber list rather than in the delivery, unlike what you meet when you buy website traffic by the session, and that difference sets it apart from most display fraud entirely. Fabricated subscribers generate impressions and clicks through automation while never existing as people, so detection has to happen downstream, at the point where a person would do something a script cannot imitate and a delivery report cannot record. Delivery reports never show the difference, and they were never built to.

Recycled bases are the more common problem in push ads inventory. A list already worked heavily by other advertisers gets resold as new. The buyer pays a fresh segment price for exhausted names, which look entirely ordinary in every delivery report the platform produces.

Signals that separate a real base from a rented one

Conversion behaviour exposes both problems faster than click data does. Real subscribers convert across a spread of delays, produce a measurable share of repeat visitors over a fortnight, and show device diversity consistent with the stated geography. Fabricated traffic converts immediately or never, and shows no repeat visitors at all. It concentrates on a narrow set of device models, because the emulation runs from a small pool of configurations. Checking device model distribution against the known market share of the target country takes ten minutes and settles the question.

Where push ads still beat cheaper formats

Despite the decay, push notifications keep a structural advantage no other cheap format offers: the message arrives while the person is doing something else, which suits offers built on impulse rather than research. Utilities, subscription trials and time limited promotions perform well against that mechanic, better than on native ads, where the visitor is already reading a page. The offer has to be comprehensible in one line.

Considered purchases perform badly here. Anything requiring comparison, a substantial payment or a long form loses most of its audience between the notification and the decision, because attention returns to whatever the person was doing within seconds. Retargeting elsewhere recovers only part of that loss.

Creative constraints reinforce that split. A title of about thirty characters and a body near seventy leaves room for one specific claim, and the icon does more work than either of them. Icons showing a recognisable object at small size beat logos, which become unreadable at the dimensions Android renders them. Emoji in the title still lift click rates in most markets, though overuse now reads as spam to the same audience it once attracted. The Adult Ad Network format table lists which platforms sell the slot at all.

The format rewards operators treating freshness as the primary purchasing variable and everything else as secondary. That ordering feels wrong to buyers arriving from search advertising, where targeting dominates every decision. It takes about one wasted month to learn. Push ads punish the older habit quickly and cheaply.