Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Metric

Push Opt-In Rate

Share of app users shown the notification permission prompt who allowed notifications.
AnalyticsmetricUpdated

Formula

Users who allowed notifications ÷ Users shown the permission prompt × 100

Unit: % · higher is better

Definition

When an app wants to send notifications, it triggers a permission dialog that belongs to the operating system. The app cannot redesign it; it can only choose when to show it. Push opt-in rate measures how that choice turned out.

Users who allowed notifications ÷ Users shown the permission prompt × 100(%)

  • Users shown the permission prompt is the number of people for whom the system dialog actually appeared in the period, usually counted per install or per user. People who never reached the screen that triggers it are outside this denominator, which is the point to remember when the timing of the prompt changes.
  • Users who allowed notifications is the number of those people who granted permission. Where some devices enable notifications by default, those users are counted separately, since they never answered a question.

Prompt-level and install-level rates

The formula above describes how well the prompt converts the people who see it. A second reading divides the same allowers by every new install instead, so it also counts the people who left before being asked. The two answer different questions, and the difference matters as soon as the prompt moves later in the journey: a delayed prompt is seen only by more engaged users, so its prompt-level rate rises almost automatically. The push permission prompt test compares timings on opt-in among all new installs for exactly that reason. Report both, and say which one a figure is.

How to interpret it

Higher is better, because the rate decides how many people any future message can reach. The push notifications and in-app messaging guide calls the permission the single most valuable asset in the channel. Read the rate in three cuts:

  • By install cohort. Group users by the week or month they installed, as a Cohort, so a change to onboarding shows up in the cohorts that received it rather than being diluted by older users.
  • By platform. iOS and Android handle notification permission differently, and older Android versions did not ask at all, so a pooled rate mixes two mechanisms. Read them apart, then together.
  • By prompt timing. If the app asks at more than one point, such as after an order or after saving a search, each trigger has its own rate and its own audience.

A rate also decays after the moment of asking. People switch notifications off in settings later, often after a run of messages they did not want. The share of active users with notifications still enabled is the measure of that decay, and a falling share is an early sign of overuse.

Illustrative example — not a real result

A bus times app moves its question

Wheelstone Bus Times asked for notification permission on the first screen after install. In one month 12,000 new users saw the prompt and 2,280 allowed it, a prompt-level rate of 19%, also 19% of installs because everyone saw it. The team moved the request to the moment a user saved a favourite stop, preceded by an in-app card explaining that alerts would warn of delays at that stop. The next month 8,400 of 12,000 new users saved a stop; 5,040 of them agreed on the card, saw the system prompt, and 4,620 allowed it. That is 92% of users who saw the system prompt and 39% of all installs, more than double the old share. Early retention by cohort held level, so the team kept the change.

What moves it

  • Timing. Asking at first launch spends the one-off prompt before the user knows what the app does. Asking at a moment of obvious value, such as an order to track or a stop to watch, gives the request a reason.
  • A soft prompt first. An in-app screen that explains what notifications are for, shown before the system dialog, lets people who are not ready decline without using up the real prompt, which can then be offered later. This changes the denominator as well as the result, which is why the install-level reading matters.
  • Honesty about promotions. If the app intends to send marketing messages, the explanation should say so; Apple's review guidelines say push must not be used for marketing unless the person opted in to it, and a prompt that hides the intent buys permissions that are quickly revoked.
  • Value shown before asking. Users who have completed a meaningful action already know why alerts would help.
  • Platform. The permission model of each operating system, and the version on the device, sets how the question is asked at all.

What good looks like

There is no reliable benchmark for push opt-in rate, and the site does not cite one. The rate depends on the category, on what notifications are for, on the platform mix and above all on when the app asks, so another app's figure says little about yours. The useful comparisons are internal: each cohort against the cohorts before it, each prompt trigger against the others, and the install-level rate against the size of push audience the app's retention plan needs. The RACE plan for app installs makes the permission rate the first Engage objective for that reason: it decides whether the re-engagement work planned after it has anyone to reach. Before changing the prompt, test the change as an A/B Test on new installs rather than comparing months, since seasonality and acquisition mix move the rate too.

Opt-in sets the reachable audience; what push does with it is measured by the engagement and retention of the people who allowed it, read by Cohort. Those messages often open a specific screen through a Deep Link, and an MMP or the app's own analytics records the opt-in event, the prompt trigger and the outcome. The deep links and app analytics guide covers the event reporting this depends on. When push is weighed against text messages, the SMS versus push notifications guide explains why push reach is capped by device permission while SMS reach is capped by consent law. The wider mobile marketing hub places both among the other app channels.

Common mistakes

  • Comparing prompt timings on the prompt-level rate, so a later prompt wins simply because fewer, keener users see it.
  • Asking on first launch and spending the only system prompt before the user has any reason to agree.
  • Pooling iOS and Android as if they used the same permission model.
  • Ignoring later opt-outs, and planning campaigns around a permission count that has quietly shrunk.
  • Hiding promotional intent in the soft prompt, which risks platform rules and lost permissions.
  • Changing the wording and the timing at once and crediting the result to the copy.

More terms to look up

Where this term is used

Related