Push notifications and retention: how to use them without annoying your customers

8 min read · Shoppy Academy

A well-crafted push notification reactivates 25% of dormant customers. A badly crafted one gets your app uninstalled within 24 hours. Between these two extremes lies the entire difference between a retention strategy that works and a chain of operational mistakes that quietly burns through the user base you fought hard to build.

If you run a Shopify store with a mobile app — or you're thinking about getting one — this article is the operating map for using push as a retention channel without turning it into a spammy megaphone. Three things will come into play: segmentation, timing, and copywriting. None of them is optional.

This article is a satellite of the pillar How to Increase Ecommerce Retention. If you want the full strategic picture, start there.

Why push notifications are the most powerful mobile retention channel

To be practical: a push notification lands on your customer's lockscreen, it's free to send, and it has an open rate that beats both email and SMS on average. It's literally the most direct way to remind a customer that you exist.

Average open rate: push vs email vs SMS

Industry data (Airship, OneSignal, Braze) is clear:

• Email: 18-22% average open rate in retail, 2-4% click-through.
• SMS: very high open rate (90%+), but significant per-message cost and very low user tolerance for a second send.
• Push notification: 7-12% average open rate on broadcast, but up to 30-45% on behaviour-triggered push.

The point isn't which channel "wins" in absolute terms — they're complementary. The point is that push is the only channel with zero marginal cost and direct lockscreen access. It's the difference between being "in the inbox" and being "in the customer's hand".

The "zero cost" of push (and why it doesn't mean free)

Sending a push costs nothing per message. But building a push system that actually works carries three hidden costs.

The first is infrastructure: you need a native app with the push SDK configured, a delivery platform (OneSignal, Braze, Firebase), and a customer data layer that lets you segment. If you run a Shopify store and you don't have an app yet, solutions like Shoppy let you turn the store into a native iOS/Android app with push enabled by default — without assembling the tech stack from scratch or hiring dedicated developers. It's the prerequisite for any serious push strategy.

The second cost is operational: someone has to think about segmentation, write the copy, schedule the campaigns, read the results. An improvised push does more damage than no push at all.

The third cost is reputational: every bad push brings the customer closer to the moment they uninstall. And a customer who has uninstalled is far harder to win back than one who never signed up.

The risk: one bad push is worth ten good ones

This is the operating principle to never forget. The user forgives one generic notification. They forgive two. On the third, they disable push. On the fourth, they uninstall. The tolerance curve is much steeper than for email, because push enters a space the user perceives as private — the lockscreen.

That means your job isn't to "send more push". It's to send fewer, but better.

The 4 types of push notifications (and when to use each)

Not all push notifications are equal. There are four categories, each with a different purpose and timing.

1. Transactional push

Order confirmations, shipping updates, delivery status, return confirmation. These are the most "permitted" push by the user: they just bought something, they want to be updated. Average open rate: 50-70%.

Don't underestimate them: a "your order is being delivered today" push that reopens the app creates the perfect moment for a well-targeted cross-sell or upsell on the homepage.

2. Promotional push

Product drops, limited editions, seasonal sales. These are the riskiest because, when sent as a generic broadcast, they produce high uninstall rates. But when well segmented by interest and purchase history, they're the ideal tool to push warm customers.

Practical rule: never more than one "broadcast" promotional push per week. For segmented promotional push, the ceiling is higher, but not above 3 per week per user.

3. Behaviour-triggered push

This is the real retention engine. They fire when the user performs (or fails to perform) a specific action: adds a product to cart and doesn't check out, adds an item to wishlist, hasn't opened the app for 14 days, completes the first purchase.

Examples of high-yield triggers:

• Abandoned cart (3-6 hours after abandonment): "Left something in your cart? We're holding it for you, today only."
• Wishlist on sale: "Product X you saved is now 20% off."
• Re-engagement: "It's been 30 days. There's something new you might like."
• Post-purchase cross-sell: "Liked X? Here are the products that go with it."

The full strategy on cart recovery via push will get its own dedicated article in the cluster.

4. Editorial and content push

The last type — the most underused. These are push notifications that don't sell directly: they announce a new blog post, a video tutorial, a community event, a "behind the scenes" drop. They keep the relationship alive even when the customer isn't in buying mode.

When the app also becomes a media outlet — not just a store — the customer opens it out of habit, not just necessity. And that's the signal retention is working. The loyalty/community angle will be covered in the dedicated Loyalty cluster pillar.

How to segment users for high open rate push

Without segmentation, any push strategy is doomed. Three segmentation dimensions to build on.

Lifecycle-stage segmentation

Three macro-segments, three message types:

• New customer (first purchase made in the last 30 days): onboarding push, product suggestions, first return coupon.
• Active customer (purchase in the last 90 days): coherent promotional push, drops, upsell.
• Dormant customer (no purchase in 90+ days): re-engagement push, personalized offer, "you missed...". Full churn-reduction mechanics will be covered in a dedicated cluster article.

Behavioural segmentation

The most effective push comes from behaviour, not demographics. Examples:

• Users who viewed the "shoes" category at least 3 times in the last 7 days.
• Users who opened the app today but didn't complete checkout.
• Users who bought product X and have high probability to want product Y (predictive cross-sell).

The more granular the segmentation, the higher the open rate. And the lower the uninstall rate.

Time-zone and time-of-day segmentation

A push sent at 11 AM Italy time lands at 5 AM in Los Angeles. It sounds obvious, but it's a common mistake even for Italy-only stores: ignoring expat users or waves of customers on vacation across different time zones.

Every serious push platform supports "smart time" delivery — sending each push at the optimal hour for each user based on their open history. Turn it on. It's a "set and forget" feature that lifts open rate by 15-25%.

Push copywriting: 5 practical rules

The difference between a push that gets ignored and one that gets opened lies in 130 well-used characters. Five operating rules.

1. Title under 40 characters, body under 90

iOS truncates titles longer than 40-50 characters, Android over 65. Body text over 90 characters gets clipped on the lockscreen. Always write for the "first glance": if it doesn't work in those characters, it doesn't work.

2. Verbs in present tense, one CTA per push

"Discover", "Shop", "Come back", "Open". No conditionals. No "you might want to...". A push is a stimulus, not advice. And every push must have one desired action. Two CTAs = no CTA.

3. Personalization with the name (when it makes sense)

"Marco, your coupon expires tonight" outperforms "Your coupon expires tonight". But use the name only when you have clean data (no "Hi undefined") and only where it really helps — not on every push, otherwise it becomes noise.

4. Emoji: yes, but only if on-brand

Emojis lift open rate by 10-25% (OneSignal data). But they have to be consistent with the brand's tone of voice. A luxury brand with exploding hearts looks fake. A DTC fashion brand using a single on-theme emoji works great.

Practical rule: one emoji per push, max two. Never at the start of the title (some devices render it poorly).

5. Real scarcity (never fake urgency)

"Today only" works — if it really is today only. If every push says "today only" and the offer actually lasts a week, the user stops believing it within three sends. Trust in push is the real asset: don't burn it for a CTR percentage point today.

Frequency and timing: the "3 in 7" rule

How many push are too many? The honest answer is: it depends on your industry and segment. But there's a rule of thumb that works for most ecommerce stores.

How many push per week before disengagement

The average tolerance threshold is 3 push per week per user. Above this, the uninstall rate grows non-linearly. Below it, you have room to test.

Recommended distribution:

• 1 transactional or service push (always permitted, doesn't count toward the "budget").
• 1 behaviour-triggered push (cart, wishlist, re-engagement).
• 1 promotional or editorial push per week for active users.

For VIP / top-spending customers you can go up to 4-5, because their tolerance is higher. For dormant users you have to come down: max 1 re-engagement push every 14 days.

The three high-open-rate time windows

Push open peaks vary by industry, but three windows almost always work in Italian retail/fashion:

• Morning, 8:00–9:30: commute, checking notifications.
• Lunch, 12:30–13:30: break, mobile scrolling.
• Evening, 20:30–22:30: post-dinner relax, the highest-volume shopping moment.

Avoid 14:00–17:00 (work focus) and after 23:00 (high risk of silent disengagement).

How to test frequency with real A/B tests

Don't pick "by feeling". Set up two user groups with different frequencies (e.g. 2/week vs 4/week) for 30 days and measure three things: open rate, conversion rate, uninstall rate. The right answer emerges on its own — and it's almost always lower than what you imagined at first.

The 3 metrics to watch every week

A push strategy without metrics is just noise. Three indicators to put on your dashboard.

1. Push open rate

The user opens the push and clicks through to the app. Industry benchmarks (retail):

• Generic broadcast: 5-8% is average, below 4% there's a problem.
• Segmented push: 12-20% is normal.
• Behaviour-triggered push: 25-40% is the target.

2. Click-through rate to product/category

Open rate is just the first step. The real indicator is CTR: how many of the users who opened the push actually clicked the deep link and completed an action (visited product, added to cart, purchased). The benchmark depends on the type, but a CTR below 30% on opens means the copy or the deep link is out of sync with expectation.

3. Post-campaign uninstall rate

The most important and most ignored KPI. How many users uninstalled the app in the 72 hours after a specific push campaign? If the rate exceeds 1% per campaign, something in the push (frequency, tone, irrelevance) is doing damage. Full churn reading will be covered in a dedicated cluster article.

Conclusion: push is a tool, not a shortcut

Push notifications can be the single most powerful lever for retention in your mobile store. They can also be the fastest way to lose the user base you spent months acquiring. Between these two outcomes there's only one thing: the care with which you use them.

Segment well. Write short. Always measure. Send less, not more. And remember that every push is a small trust contract with the customer — a contract you can renew every day or break with a single bad notification.

A well-crafted push is worth more than a thousand emails. But only if it's well crafted.