How often should a store send push notifications?

By Abdalla Kamesh · Published September 26, 2026 · 4 min read

Push notification permission is the only marketing asset you can lose permanently in a single tap. Email addresses can be re-collected. A customer who turns off your notifications — or worse, uninstalls — is not coming back to Settings to turn them on again.

That asymmetry should govern how you think about frequency, and it usually doesn’t.

Two budgets, not one

The most useful mental shift: transactional and marketing notifications draw from different budgets.

Transactional notifications do not count against your frequency budget. Order shipped, out for delivery, delivered, refunded, payment failed. Customers do not experience these as marketing — they experience them as the store working. Send all of them. The full list of which statuses to notify on.

Marketing notifications do count. Campaigns, sales, new arrivals, re-engagement. This is the budget with a limit.

Triggered notifications are in between. Back-in-stock and abandoned cart are marketing by intent and transactional by feel, because the customer’s own action caused them. They are the best value in the whole channel: high relevance, low irritation. Count them at half weight.

The cap

For most stores: two to three marketing notifications a week, maximum.

Daily works only when daily is the product — flash-sale retail, groceries, anything where new inventory is the reason people installed. If you are a clothing brand releasing a collection a month, three campaigns a week means you are inventing reasons to interrupt.

Set the number in advance and write it down, because in the moment “one more send” always looks free. It is not; it is drawn from an account you cannot top up.

Segment instead of increasing frequency

The rule that resolves most frequency arguments: sending twice as often to half as many people beats doubling the blast.

You have an order history and a wishlist inside the app. Use them:

Bought outerwear → the outerwear restock, not the whole sale.
Wishlist item back in stock → that person only.
Has not opened in 60 days → one re-engagement message, then leave them alone.
Bought in the last week → skip the campaign entirely; they are already customers this month.

Every one of these is a smaller send with a higher tap rate and a lower opt-out cost. That is the trade you want, every time.

Quiet hours

Never between about 10pm and 8am in the customer’s local time. Not yours.

This is not politeness, it is retention arithmetic. A notification that wakes someone is not evaluated on its content; it is evaluated on having woken them, and the resolution is usually to turn notifications off.

The corollary: send campaigns in the customer’s local evening, when people browse. A campaign sent at 9am your time to customers in another timezone is both badly timed and possibly nocturnal.

The two numbers that tell you when you went too far

1. Opt-out rate per campaign. Notifications disabled, divided by delivered. Track it per send, not monthly, so you can see which message did it. A single campaign with a visible spike is a message worth examining.

2. Uninstalls in the 48 hours after a send. The harsher signal, and the one people ignore because it is unpleasant. Uninstalls are permanent.

Watch both against your own baseline rather than a benchmark. And note the trap: open rate frequently rises just before an audience leaves, because the people who remain are the engaged ones — the audience is shrinking while the ratio improves. Open rate alone will tell you a heavy week went well.

Recoverable and unrecoverable damage

Recoverable: a campaign that flopped. A week with one send too many. A message that was less relevant than you thought. Cut back, send something genuinely useful, and the audience recovers.

Unrecoverable: disabled notifications, uninstalls, and the customer who now associates your brand with interruption. There is no re-permission flow.

Which is why the correct posture is conservative. The cost of sending one campaign fewer is small and reversible. The cost of one too many is neither.

A weekly rhythm that works

For a store with a normal product cadence:

Monday: nothing. People are not shopping.
Wednesday evening: one campaign — new arrivals, or a genuine reason.
Weekend, if you have something real: one more.
Any day: triggered notifications, unlimited by definition because they are one-to-one.
Always: transactional, unlimited.

That is two marketing sends, plus triggers, plus transactional. Sustainable indefinitely.

Before you send anything, five questions

Would I want this on my own lock screen at this hour?
Is this the second message this week, or the fourth?
Could this be a segment instead of a blast?
Does the tap land on the exact screen it promises?
What am I willing to accept as an opt-out rate for this?

Question 5 is the discipline. Deciding the acceptable cost before you send is how you find out whether the campaign was worth it — rather than deciding afterwards, when the answer is obvious and the permission is gone.

Related reading

In StoreToNative, the Analytics page shows push opens against pushes sent, installs and app revenue for the period you pick, and Recent sends on the Push page lists every send and how many devices it reached. Opens and revenue are totals for the app, not split per campaign, so leave room between sends when you want to tell one from another.