What is a push notification?

Why the channel is valuable and fragile

Push reaches a user who is not currently in your product, on a surface they check constantly, at essentially zero marginal cost. Nothing else in mobile marketing has that combination.

It is also the most easily destroyed channel available, because the user holds a single irreversible control: turning notifications off. Once revoked, permission is very rarely granted again, and the user is unreachable through the channel permanently.

That asymmetry is the whole discipline. The cost of a bad notification is not a low open rate; it is the option value of every future notification to that person.

What works

• Ask for permission at a moment of value, not on first launch. A prompt before the user understands the product is a wasted request.
• Send about the user, not about you. Their order, their match, their reminder — not your campaign.
• Respect time zones and quiet hours, which is basic and frequently not done.
• Cap frequency centrally, so several teams cannot each send their own reasonable notification on the same day.
• Measure opt-out rate as the primary metric, not open rate. Open rate optimises for the next send; opt-out rate protects the channel.

Push and re-engagement measurement

Push-driven returns are easy to over-claim, for the same structural reason as reattribution: notifications go to people who already use the product, many of whom were returning anyway.

A notification sent to an engaged user and a return an hour later is not evidence of causation. The instrument that separates prompted returns from spontaneous ones is a holdout against a matched segment — see incrementality — and the results routinely surprise teams reading open-rate reports.