Last updated: 8 September 2026

The targeting settings that decide whether push ads pay for themselves

Every network exposes the same short list of filters, and almost nobody sets more than half of them before a first campaign goes live. Push ads targeting works less like an aiming device and more like a series of exclusions, each one removing a slice of traffic that would otherwise burn budget without ever converting. Device, carrier, connection type and browser version each catch a different failure mode, and skipping any one of them means paying full price for impressions the offer was never going to accept in the first place.

Device and OS filters at the centre of push ads targeting

Classic subscriber based push runs almost entirely on Android, since Apple never allowed a browser prompt to register a subscription the way Chrome and Firefox do. Push ads targeting an offer without a device filter therefore sends a meaningful share of budget toward iOS traffic that the delivery mechanism cannot actually reach at all, and the wasted spend shows up as unusually low volume rather than as an obvious error anywhere in the dashboard.

In-page push exists specifically to close that gap, rendering a notification style unit inside the page itself rather than through the operating system, and it works on every platform a browser runs on. Buying it as a substitute for iOS coverage rather than as an extension of Android coverage keeps the two delivery types from being blended into one misleading average.

Operating system version matters almost as much as the platform itself, because permission prompts changed meaningfully across major Chrome and Android releases, and a subscriber base collected under an older prompt style behaves differently from one collected after the interface tightened.

Why Android and iOS need separate campaigns, not one split rule

A single blended report across Android and iOS in-page inventory hides both halves equally well, which is the same problem that shows up in geo blending and in segment age blending elsewhere in this format. Splitting the buy into two campaigns costs one extra setup step and answers a question that a combined report will never settle on its own terms.

Desktop and mobile also deserve separate treatment inside push ads targeting even on the same operating system, since notification behaviour, screen context and the moment of interruption differ enough between the two that a single creative rarely performs identically across both.

Carrier and connection filters inside push ads targeting

Carrier level targeting exists on most networks and is used by a small minority of buyers, largely because the setting sits several menus deep and produces no obvious symptom when it is left open. Push ads targeting without a carrier filter allows delivery through connections known for compromised devices and simulator traffic, and the resulting clicks look identical to genuine ones in a basic report.

Excluding known problem carriers and data centre ranges before launch removes a category of traffic that almost never converts and almost always registers a click, which is precisely the combination that inflates a click through rate while quietly destroying the return on spend behind it.

Wifi and mobile data behave differently enough to warrant separate bids on any offer sensitive to page load speed, since a subscriber notified while on a slow mobile connection abandons a heavy landing page at a materially higher rate than one browsing on wifi, and the gap widens further on markets where mobile data remains genuinely expensive per megabyte.

Wifi against mobile data and why the split changes conversion

A campaign running one bid across both connection types is implicitly assuming they convert the same, and that assumption rarely survives contact with a real report once volume is high enough to split reliably. Testing the split costs a single extra campaign and usually pays for itself inside the first week once the weaker connection type is capped or excluded outright.

Browser version filters catch a smaller but real category of loss, since notification rendering changed across major releases and a subscriber stuck on an old build sometimes cannot see a creative element the rest of the audience takes for granted.

None of these filters require more than a checkbox and a saved preset once configured, yet the account managers who mention them unprompted are noticeably rarer than the ones happy to discuss creative and headline testing instead, which says something about which lever actually moves the number and which one is simply easier to talk about in a sales call.

Geo and language layers that refine push ads targeting further

Country level targeting is the first filter every buyer sets and the one requiring the least explanation, yet push ads sold at the country level still mix language communities that respond to entirely different creative, particularly in markets with more than one dominant first language spoken at scale.

Push ads targeting refined to a language or region layer beneath the country filter usually costs a small amount of reach and returns a disproportionate improvement in response, because a notification written in the majority language reads as noise to a subscriber who does not use it day to day, regardless of how strong the offer underneath it might otherwise be.

Regional splits inside a single country matter most in markets with genuinely separate metro pricing tiers, where an offer viable in a capital city returns nothing at all once delivered to a rural region with a different average income and a different relationship to the product category entirely.

Language mismatch as the quiet cause of a weak click through rate

push notification ads that ignore time zone inside a large country deliver at the wrong local hour for a meaningful share of the audience, since a single send window applied nationally treats a five hour spread as if it were a single moment, and the subscribers reached at three in the morning local time behave nothing like the ones reached during the evening.

Scheduling by local time rather than by the buyer's own time zone is a setting available on most platforms and switched off by default on nearly all of them, which makes it one of the highest value adjustments available before a single dollar of extra bid gets added anywhere.

Regional pricing tiers inside one country rarely appear anywhere on a rate card, since the platform sells at the national level and leaves the buyer to discover the internal split through their own reporting, one region cut at a time across a handful of test weeks.

Building a push ads targeting checklist before the first launch

Order matters more than most buyers assume, because push ads targeting applied all at once makes it impossible to attribute an improvement to any single filter, and a campaign that suddenly performs better after five changes teaches nothing about which of the five actually mattered.

The order settings should be applied, not just the list of them

Device and OS exclusion comes first, since it removes traffic the delivery mechanism cannot reach at all. Carrier and connection filters come second, removing traffic that reaches the device but rarely converts once it lands. Language and regional layers come third, refining an already clean audience rather than trying to fix a fundamentally mismatched one. Local time scheduling comes last, because it only matters once the underlying audience is already worth reaching at the right hour.

Applying the list in that order and measuring after each step turns a single launch into four small experiments instead of one large guess, and the record from each step carries forward into every future campaign run on the same offer or a similar one.

Keeping that record in a single shared sheet rather than in each manager's head is the part most teams skip under deadline pressure, and it is also the part that turns a one off launch checklist into something the next campaign can actually reuse instead of rebuilding from memory.

A launch checklist I built for a client leaned on the targeting documentation at push-ads.io, and the detail that earned its place in the final version was how plainly the platform separated reach limiting filters from response improving ones, which is exactly the distinction that turns a vague targeting menu into an ordered sequence a team can actually follow without arguing over which setting to touch first.

Push ads targeting done properly removes far more waste than any single creative change will manage, because a filter excludes bad traffic permanently while a stronger headline only ever competes for attention against an audience that was never going to convert to begin with.