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

Mobile & App Marketing › Troubleshooting

Push Notification Opt-In Rate Low

Why few app users allow push notifications and how to fix it: a prompt at first launch, no soft prompt, vague wording, one switch and a misread rate.
Troubleshootingintermediate9 min readUpdated

First, confirm what the rate is measuring

Write down the denominator before reading the number. The push opt-in rate entry describes two readings of the same allowers: divided by the people who saw the prompt, it says how well the prompt converts; divided by every new install, it also counts the people who left before they were asked. A prompt that moved later in the journey looks better on the first reading almost automatically, because only more engaged people reach it, so a team comparing the two readings across a change can see a rise or a fall that never happened.

Then split by platform. The push notifications and in-app messaging guide explains that on iOS an app must ask before sending notifications and the system prompt can be shown only once, and that recent Android versions have moved to the same model with a runtime permission. The opt-in rate entry adds that where some devices enable notifications by default, those users belong in a separate count, since they never answered a question. A blended figure moves whenever the mix of devices moves. Finally, read the rate by install Cohort, so that a change to onboarding shows up in the weeks that received it instead of being diluted by years of older users. With the measurement settled, the table below separates the real causes.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
The dialog appears before the first useful screen and most people declineWhere the prompt fires in the first-session funnel, screen by screenSystem prompt shown at first launch, before the app has shown any valueMove the request to the moment notifications are plainly useful
The system dialog is the first and only time the app mentions notificationsWhether any in-app explanation precedes the system promptNo soft prompt, so the single system prompt is spent on people who would refuseAdd a soft prompt and show the system dialog only to people who agree
A soft prompt exists, but few agree to itIts wording against the messages the app actually sends, including promotionsA soft prompt that is vague about what will be sent, or silent about promotionsName the messages in the person's terms and say plainly if offers will follow
People who would want order or booking updates still declineWhether settings let people choose kinds of message, and what the soft prompt offersOne switch for every kind of message, so allowing means accepting all of itOffer topics, with service messages kept apart from promotions that start off
The rate moved after a change, or differs sharply by reportThe denominator, the platform split and the cohort behind each figureThe rate read on a shifting denominator or blended across operating systemsReport install-level and prompt-level rates, per platform, by cohort

Fix, verify and prevent each cause

A request at first launch

A person who has just installed an app does not yet know what it does, so a request to interrupt them later is a request to trust a stranger. The guide to app onboarding and permission requests treats the first session as a funnel from install to the activation moment, and lists notification permission among the things usually deferred until notifications are plainly useful: after an order someone will want to track, a search they want alerts on, a booking they could forget.

Fix: map every screen between first open and the activation moment, remove the permission request from launch, and attach it to the first action that makes the value obvious.

Verify: among new installs, counted across every install rather than only those who reached the prompt, the share with notifications allowed rises in the cohorts that received the change, and early retention does not fall.

Prevent: keep the first-session screen map in the release checklist, so that a new prompt cannot be added at launch without someone deciding where it belongs.

No soft prompt in front of the dialog

Because the system prompt can be shown only once on iOS, showing it to everyone wastes it on the people who would have said no. The push guide sets out the pattern that protects it: an ordinary in-app screen or card explains what notifications will do, and only people who agree there see the system dialog; people who decline have not used the system prompt, which can be offered again at a later moment of value.

Fix: build the soft prompt as part of the app, with agreeing and declining equally easy, and trigger the system dialog only from the agree button.

Verify: the share of people who see the system dialog falls, the share of those who allow it rises sharply, and the install-level rate rises or holds.

Prevent: log the soft prompt shown, its answer, the system prompt shown and its answer as separate events, so a future change that bypasses the soft prompt shows up in the event counts the same week.

A soft prompt that does not say what will be sent

A soft prompt earns trust only if it is specific and honest. The onboarding guide asks for the messages to be named in the person's terms, such as a delivery on its way, not the app's own phrases about staying in the loop, and says the prompt should never imitate the system dialog or make the app look broken until the person agrees. It also records that Apple's App Store Review Guidelines say push must not be required for an app to function and must not be used for marketing unless the person has opted in, so a soft prompt that is quiet about offers the app intends to send is a problem twice over.

Fix: rewrite the soft prompt around the two or three messages a new user will actually receive first, and state separately whether promotional messages will follow and how to choose them.

Verify: agreement with the soft prompt rises without a later rise in people switching notifications off.

Prevent: have the owner of the message programme approve the soft prompt wording whenever a new message type is added, so the promise and the programme stay the same.

One switch for everything

If allowing notifications means accepting order updates and every promotion together, some people who want the updates refuse the whole bundle. The tutorial on notification preferences by topic builds the alternative: an inventory of every message each team sends, topics people recognise, transactional messages kept apart from promotional ones, and promotional topics off unless the person chose them.

Fix: give the app a settings screen with topics, make every sending tool read it, and let the soft prompt point to it so the person knows offers are a separate choice.

Verify: the soft prompt's agreement rate rises, and topic choices show people keeping service messages while declining offers rather than refusing everything.

Prevent: name an owner for each topic who approves any new message into it, as the tutorial recommends, so offers do not creep into the topic everyone keeps on.

A rate that is not measuring what the team thinks

Sometimes nothing is wrong with the prompt. The rate was divided by people who saw it in one report and by every install in another, or an Android figure inflated by devices enabled by default was averaged with an iOS figure. The push permission prompt test reads opt-in among every install assigned to each arm, splits iOS and Android before pooling them, and treats early retention as the guardrail, for exactly these reasons.

Fix: define both rates in writing, report them per platform and per install cohort, and keep devices that never answered a question in their own count.

Verify: the figures in every report reconcile to the same event counts, and a change in device mix no longer moves the headline.

Prevent: store the definition beside the dashboard and change it only with a note in the measurement change log.

A prevention routine for the permission

The permission is the ceiling on every push message the app will ever send, so it deserves a standing check rather than a fix after a bad quarter. Make the install-level opt-in rate, per platform, a named KPI for the messaging programme, read weekly by install cohort beside activation and early retention. Before any release that touches onboarding, walk the first-session screen map and confirm where each permission is requested, that the soft prompt still precedes the notification dialog, and that its wording still matches the messages the app sends.

Before any new message type launches, place it in a topic and check that the soft prompt and the settings screen still describe the programme honestly. When a team wants to move the prompt or change its framing, run it as a test rather than a before-and-after comparison; the push permission prompt test sets out the arms, the denominator and the guardrails. And watch the other end of the funnel: the push guide treats a falling share of users with notifications switched on as the earliest sign that the channel is being overused, which is a problem for the message programme rather than the prompt, and is listed with the other app symptoms in the app marketing troubleshooting section.

Illustrative example — not a real result

Wimblefern Fixtures moves its prompt behind a followed team

Wimblefern Fixtures, an invented app for an amateur football league, sends fixture changes, kick-off times and results, plus occasional offers from league sponsors. Its dashboard shows a push opt-in rate of 52%, which the team calls low. Split by platform, iOS shows 34% of people asked allowing notifications at first launch; the Android figure of 71% mixes devices that asked with older devices enabled by default, which never answered a question. Measured across every new install, iOS opt-in is 31%. The team removes the first-launch prompt and shows a soft prompt after a user follows a team: "We'll tell you when a fixture moves, when kick-off changes and when the result is in. Sponsor offers are a separate switch and stay off unless you turn them on." A settings screen offers three topics. Over four weekly cohorts, half of new iOS installs get the new flow. Opt-in across every install reaches 43% against 30% for the old flow; among people shown the system dialog it is 88%, a figure the team reports but does not headline. Day-14 retention is within a point of the old flow. Of those who allowed notifications, 9% switch sponsor offers on.

Common mistakes

Rewriting the soft prompt when the real fault is that it appears at launch. Comparing a prompt-level rate after a change with an install-level rate before it. Averaging iOS with Android devices that never asked. Hiding promotions inside a prompt about service updates. Asking again and again after the system prompt was declined.

Frequently asked questions

What is a good push opt-in rate?

There is no figure worth borrowing, because the rate depends on the app's category, its audience, each operating system's permission model and, above all, on who is in the denominator. Compare the app with its own history instead: read iOS and Android separately, read by install cohort, and report the rate among every install as well as among the people who saw the prompt. A change that raises one and lowers the other has not yet proved itself.

Can we ask again after someone says no?

It depends on which question they declined. If they declined the app's own soft prompt, the system prompt has not been used and can be offered later at another moment of obvious value. If they declined the system prompt, the push guide explains that on iOS the only way back is through the device settings, so repeated asking achieves nothing. Make sure the app works well without notifications, as Apple's guidelines require, and let in-app messages carry what matters during sessions.

Can we change the wording of the system prompt?

No. The permission dialog belongs to the operating system, and the push opt-in rate entry notes that the app can only choose when to show it. Everything a team can improve sits around the dialog: the moment it appears, the soft prompt that comes first, what that soft prompt promises, and the preferences that let people choose which kinds of message they receive.

Terms used on this page

Where this page is used

Also related

Sources

  1. App Store Review Guidelines — Apple Developer

Related