Why delivery timing decides outcomes for push ads

Written by Smith Jones, Performance Media Buyer/Analyst · Last updated 26 August 2026

The format exists because somebody once tapped Allow on a browser prompt and then forgot the whole thing. Everything that follows depends on that one permission, on how long it survives, and on how many advertisers are queued behind you waiting to use it. Anyone buying push ads is renting access to a list assembled by somebody else, which means the age of that list, the exact moment of delivery, and the wording inside a very small box all matter far more than the bid attached to the campaign itself.

Subscription flow that produces push ads

A visitor grants permission to a single domain. Nothing in the chain belongs to the advertiser, since the publisher holds the subscriber while the network holds the commercial relationship sitting behind every campaign built on push ads.

The browser then registers a background script for that domain. That script keeps the subscription alive and can wake a device long after the site asking for permission has been forgotten entirely, because delivery happens at the level of the operating system rather than inside a browser tab, which is why platform account structures built for display map onto this format poorly. That placement is the proposition.

Prompt placement and the cost of asking too early

A prompt fired the instant a page loads gets refused by most visitors, and refusal on that domain is close to permanent because nobody reopens browser settings. Publishers who care about their inventory show a custom invitation first, so a declined offer costs nothing at all and can be repeated on a later visit. Asking after a visitor has watched something, downloaded something, or reached a second page produces opt in rates several times higher than asking at the door, while a visitor declining the prompt twice is gone for good and the inventory shrinks by one.

The subscriber belongs to a publisher, so an advertiser buys access. Networks aggregate subscriber pools from thousands of separate sites, the publisher terms for which are documented on Adult Ad Networks, and resell reach across all of them, which is why one person receives messages from unrelated brands inside an afternoon, and why nobody selling this inventory offers exclusivity. Plans do not change it.

List decay and what it removes from push ads

A subscriber base loses value during every week that it continues to exist. People unsubscribe deliberately, clear browser storage, replace a handset, switch browsers, or simply stop reacting to anything that arrives, and every one of those events quietly removes a person from the audience while leaving the count on the invoice completely untouched. The list still looks the same size on the invoice as it did back in January. Decay of that kind sets the ceiling on what any account can spend productively on push ads, whatever the reporting suggests.

Invisible. Delivery attempts still succeed technically, impressions still register, and nobody at all is looking at the screen where any of them land. Click rate falls while volume holds perfectly steady. That is the signature, and it shows for months before anybody investigates.

Subscriber age as a targeting parameter

Age is the strongest single predictor in the account, and most networks expose it in days without being asked. Somebody who granted permission this week reacts at several times the rate of somebody who granted it eight months ago. I checked which networks expose that parameter against the comparison kept on push-ads.io, since a platform hiding subscription age forces one blended price on everybody. Bidding one figure across every age band overpays for the dormant part of the pool and loses the fresh part to competitors.

Fresh inventory carries a hard ceiling on it. An account scaling on new subscribers alone reaches that ceiling inside a few weeks and then keeps buying nothing but silence, since recruitment speed depends entirely on publisher traffic and no budget accelerates it.

The workable structure splits campaigns by age band, keeping separate budgets and separate creatives running against every one of them. Older cohorts still convert on offers requiring no urgency and cost a fraction of the fresh tier, so judging both on cost per action rather than click rate keeps the cheap volume inside the account instead of discarding it after one impatient week. Deletion is permanent.

Elements inside a working creative for push ads

The unit assembles from an icon, a larger image, a title, a description, and on some systems a pair of action buttons sitting underneath it. Every one of those slots is optional in the specification and close to mandatory in practice, because a message arriving without an image occupies half the height of one that includes it. Rendering happens inside the operating system, so the preview shown in a network interface is a drawing instead of a result, and the difference between those two is where most disappointing campaigns for push ads begin.

The icon is the only element guaranteed to display in every environment, unlike adult inventory pricing where the image carries most of the weight. Decoration wastes it. A logo shrunk to thumbnail size reads as a grey square on a small screen, while a single letter or a recognisable silhouette stays legible at any dimension a system decides to render it.

Icons, titles, and where the system truncates

Titles disappear into an ellipsis at roughly thirty characters on a handset, while descriptions collapse to a single line until the recipient chooses to expand the notification. Most never expand anything at all, so that second line ends up written for nobody. Loading the whole proposition into the opening words keeps it safe from every cut a system makes, whatever a platform claims about its own limits. Test on a real handset, because two minutes there settles every argument about character counts.

Misleading titles buy clicks and then pay for them twice over. Publishers watch unsubscribe rates closely enough to drop an advertiser outright, and reputation inside a network remains the one asset an advertiser can never simply purchase back later.

In page formats sitting beside push ads

A second family of units imitates the same appearance without any permission. These render inside the page as a notification styled block, reach visitors on systems where subscriptions never worked, and disappear the moment a tab closes, so nobody subscribed to anything, nobody can unsubscribe, and the audience regenerates with every page view a publisher serves. Buyers reach for them when volume matters more than persistence, and they behave nothing like push ads once anybody measures them.

Volume differs in both directions here. In page inventory is larger because every single visitor qualifies for it and weaker on website traffic quality because attention has already gone to the page underneath, so pricing follows that logic and the cheaper unit is cheaper for a reason worth understanding first.

Delivery windows and the queue behind them

A message sent to a device that is switched off waits in a queue, and a time to live setting decides whether it arrives or expires unseen. Expiry matters. An offer valid for one evening, dispatched with a three day setting, reaches part of its audience fully two days after the reason for sending it has gone. In page units carry no queue at all, since the impression either happens while somebody is reading or never happens, which makes them the safer choice for anything time bound. Matching expiry to the offer is a two second decision nobody makes.

Reporting shapes differ. One format separates messages sent from messages actually shown, unlike porn traffic slots where an impression either rendered or did not, and that gap is where dormant subscribers hide. The other reports impressions alone, since a rendered block is the whole event. Comparing cost per click across both produces a false conclusion.

Frequency capping applied to push ads

Caps belong at the level of one subscriber across one day, and the number that actually works is a good deal smaller than most accounts assume before they measure it. Three arrivals in a day sit at the edge of tolerance. Beyond that the extra impressions produce unsubscribes at a rate outrunning whatever they earn. Every unsubscribe removes a person permanently, so overexposure shrinks tomorrow audience to pay for clicks today, which is the arithmetic that governs profitable spending on push ads.

Two. Yours limits what a single advertiser sends, while the network limits what one subscriber receives from every advertiser on the platform at once.

Send timing and the rotation behind it

A generous network cap means competitors stand in that same queue, while a strict one means the third message of the day never leaves the building, and neither number appears in any interface. Send timing decides a surprising share of the result on top of that. Local evening hours outperform working hours on consumer offers, and a campaign scheduled in the buyer timezone rather than the recipient timezone arrives at three in the morning for half the audience. The control is buried somewhere different on every platform, under a different name, and the default is seldom what anybody actually wants. Checking costs a minute.

Creative rotation finishes what capping starts. A subscriber who has seen the same icon and headline four times over has stopped reading either, and swapping the image recovers part of the response while swapping the title recovers more. Both. Keeping three approved variants in rotation and dropping each as soon as its numbers flatten turns an account into a slow continuous test, which remains the only durable way anybody has found to carry on spending on push ads.