Klaviyo Push Notifications: When They Earn Their Place

TL;DR. Klaviyo push notifications require a mobile app you own, with the SDK embedded. That prerequisite rules out most e-commerce brands. Where the app exists, push has a real advantage, near-zero marginal cost, unlike SMS, and a specific risk: a saturated contact does not unsubscribe, they uninstall. The penalty is heavier and irreversible.

Push is regularly presented as a free alternative to SMS. The comparison misleads on two counts: it reaches only a fraction of your list, and its penalty for over-sending is of a different nature.

The prerequisite that settles it

Klaviyo sends push notifications through a mobile app you own, with the Klaviyo SDK embedded and system permissions granted by the user.

In other words: no app, no channel. This is not web push, which runs on different mechanics and different tools.

That prerequisite has a practical consequence worth accepting early: push does not address your list, it addresses your app holders. That is usually a small share by volume and a high one by value : your best customers.

The asymmetry shapes everything else. A channel that only speaks to your best customers should not be used to push generic promotion.

What push does better

Cost. Unlike SMS, where every send is billed, the marginal cost of a notification is near zero. That changes the arbitration on low-basket flows where SMS is not profitable : see SMS marketing cost.

Immediacy without perceived intrusion. The notification arrives on a channel the user explicitly installed. The perception differs from a text.

In-app context. The tap returns the user to the app, where the purchase journey is usually better optimised than on mobile web.

Use cases that work

Back in stock. As in SMS, the best case: the message is expected and availability is limited.

Price alerts. On catalogues where price moves, with product-level opt-in.

Order updates. They replace transactional SMS for app holders at zero cost.

Cart reminders, complementing and never duplicating the email. The flow logic is in Klaviyo abandonment flows.

What does not work

Content and catalogue. The space allows one message and one action. Anything requiring you to show belongs in email.

The generic promotional campaign. This is the use that produces uninstalls, and it targets precisely your best customers.

Duplication. A contact receiving the email, the text and the push on the same offer has received three solicitations. The non-overlap rule applies across all three : see SMS or email and Klaviyo omnichannel marketing.

The specific risk: uninstalling

This is what separates push from every other channel.

A contact saturated by email unsubscribes: you lose a recipient. A contact saturated by push deletes the app: you lose the channel, the in-app browsing history, saved preferences, and often the customer.

The penalty is heavier, and irreversible without reacquisition : which costs considerably more than an email re-subscribe.

The steering consequence is that you must track uninstall rate by send cohort, not only notification opt-out rate. It is the only metric that reveals a campaign's real cost, and it appears in no revenue report.

The channel nobody budgets for, and why that matters

Push has a property no other owned channel has: the marginal cost of a message is close to zero, and the cost of a mistake is close to total. An email that annoys produces an unsubscribe from one channel. A push that annoys produces an uninstall, and you lose the app, the account and every channel attached to it at once.

That asymmetry should drive the plan rather than the volume. The right question before each send is not what it will bring, it is what it costs if the recipient did not want it.

It also explains why push rewards precision more than reach. A notification tied to something the person asked for, a restock, a delivery, an expiring basket, is welcomed. The same notification sent broadly is the one that gets the app deleted.

Frequency

Whatever cap you set, write it down and hold it across the year. Push is the channel where a single busy fortnight can undo a base built over months, and the damage does not show in the campaign report: it shows later, as a smaller addressable audience with no visible cause.

Lower than intuition suggests, precisely because the penalty is heavy and the target is your best customers.

Our position: push stays a flow channel, not a campaign channel. Push campaigns earn their place on rare events : a major launch, a private sale reserved for app holders, and no more than once or twice a month.

What to check before opening the channel

Three things decide whether push is worth building, and none is about the message.

Opt-in rate at install. The system prompt is asked once and answered fast. An app that fires it on first launch, before showing any value, converts far below one that waits for a moment where the reason is obvious, a restock request or an order to track.

The share of your customers who have the app at all. Push reaches app users, not customers. A brand where the app carries a small fraction of orders is building a channel for a minority, and the same effort spent on email segmentation would touch everyone.

What happens when someone declines. On most platforms the prompt does not come back, so a refusal is effectively permanent. That single fact is what makes the timing of the request more important than the wording of any notification you will send afterwards.

How to measure

Revenue per recipient, as on other channels. Our reference framework puts the email alert threshold at €0.10 per recipient; push has no routing cost, so the break-even is lower, but the cost of uninstalls raises it invisibly.

Uninstall rate by cohort, covered above.

App open rate by entry channel, which says whether push actually drives traffic or merely notifies.

FAQ

When should the permission prompt be shown?

At a moment where the reason is obvious to the person, not on first launch. On most platforms the prompt is asked once and does not return, so a refusal is effectively permanent. A restock request or an order to track converts far better than a prompt shown before any value has been demonstrated.

How does push compare to email in risk?

The costs are asymmetric. An email that annoys costs an unsubscribe from one channel; a notification that annoys can cost an uninstall, and with it the app, the account and every channel attached to it. Plan on the cost of being unwanted, not on the upside.

Do we need a mobile app to use Klaviyo push?

Yes, with the SDK embedded. Without an app, the channel does not exist.

Does push replace SMS?

On some messages yes, at near-zero marginal cost, but it only reaches app holders.

What is the real risk with push?

Uninstalling, which costs you the channel, the history and often the customer.

Which use cases work?

Back in stock, price alerts, order updates, cart reminders.

How do we measure a push programme?

Revenue per recipient, and above all uninstall rate by send cohort.

Going further

Channel arbitration is covered in SMS or email, and flow architecture on our marketing automation agency page.

Review your email and CRM stack with Deliver