Apple and Google now control the entire push‑notification pipeline, meaning every alert a brand sends passes through one of their services. Since the launch of APNs in 2009 and Google’s Cloud Messaging evolution, the two platforms have acted as intermediaries, capable of throttling, dropping or reordering messages. Their growing intervention reshapes how marketers reach users.

Android 8’s 2017 Oreo release introduced notification channels, forcing developers to predefine categories and importance levels, while iOS 15 added Focus and a four‑tier interruption taxonomy in 2021. A year later Android 13 required a runtime POST_NOTIFICATIONS permission, slashing opt‑in rates; Pushwoosh recorded a 30% drop for games and a 19% dip for news apps.

Both ecosystems now apply on‑device AI summarisation to summarise, reorder or even rewrite notifications, leaving marketers with little visibility—no inbox‑style archive or deliverability dashboard exists for push. As platforms treat attention as a guarded asset, senders must adapt to opaque, model‑driven filtering rather than rely on direct delivery. The reality is that push messaging has become a controlled channel.

Developers seeking reliable reach now turn to analytics SDKs, custom token management and A/B testing of notification payloads to gauge platform behavior. Some vendors experiment with server‑side throttling and priority tagging to stay above algorithmic thresholds. Ultimately, success hinges on designing concise, time‑sensitive alerts that align with the platforms’ AI preferences.