A 45% push open rate looks great in a slide until you ask how many people actually said yes. If only 12% of installers opted in, and half of those never open anything useful, you do not have a retention channel. You have a small, noisy list with a flattering numerator.
This piece is for ecommerce directors and CRM leads who want an ecommerce push notifications app channel they can defend in a weekly meeting. Not blast volume. Permission quality. When to ask, how to frame the value exchange, how to protect the channel after opt-in, and which numbers to put on the dashboard.
At ConvertNative we ship fully managed native iOS and Android apps on subscription for brands on Shopify, WooCommerce, PrestaShop, and Magento. Push only pays when the permission pool is real and the product path behind the tap is clean. For open-rate math once you already have opt-ins, see push open rates vs email. For cart-level sequencing, see abandoned cart recovery: email vs push vs in-app.
Why permission rate is the real lever
Open rate measures people who already said yes. Permission rate measures how many of your installers (and later, your buyers) ever enter that pool.
Two stores can both report "strong push opens" and live in different realities:
Store A: 40% of installers opt in. Segmented sends. Opt-outs stay flat. Push-attributed orders show up every week.
Store B: 10% opt in after an on-first-open nag. Weekly blasts to everyone. Opens look fine on a tiny base. Opt-outs climb. CRM stops trusting the channel.
Finance and growth should care about Store A's shape. Store B is a demo metric.
Push still beats email on attention when the cohort is clean. ConvertNative programs see push opens around 45%, about 7x typical email, and roughly -20% reactivation cost versus ads and email alone. Those figures only transfer if you built a permission base worth messaging. Method first. Brag second.
When to ask (and when to shut up)
iOS and Android both punish bad timing. Ask too early and you train a No that is hard to reverse. Ask with no context and you look like every other app that wanted the lock screen on second one.
Post-value moments beat first-open nags
Ask after the shopper already got something from the app. Examples that work in ecommerce:
First successful order confirmation inside the app
Wishlist or back-in-stock save
Loyalty tier unlock or points balance they can see
A browse session that clearly shows catalog speed or saved cart state
The pattern is simple. They felt a reason the brand belongs on the phone. Then you explain what push will do for them. Then the system dialog.
Skip the second-one OS prompt
Triggering the native permission dialog on first launch is still common in DIY templates. It is also how you manufacture permanent refusals. Soft-ask first. Native dialog second. Never reverse that order unless you like living with a blocked cohort.
Quiet periods after a No
If they dismiss the soft ask, wait. Change the context (post-purchase, restock interest) before you try again. Re-prompting every session is how you get 1-star reviews and support tickets, not a bigger list.
Pre-permission UX and an honest value exchange
The pre-permission screen is not decoration. It is the only place you control the story before the OS takes over.
Say what you will send
Write the bargain in plain language:
Order and shipping updates
Back-in-stock on items they saved
Early access or member drops they can opt into by category
Cart reminders when they leave something behind
Do not promise "tips and offers" if the real plan is three promos a week. People remember the lie when the third discount lands at dinner.
Say what you will not do
One line of restraint builds more trust than five feature bullets. Example: no more than X marketing pushes per week, no overnight noise in their timezone, easy mute by topic later. Then keep the promise in product settings and in your send tooling.
Design the soft ask like a product surface
Use brand UI, one primary CTA ("Enable alerts"), one clear dismiss. Show a mock notification that looks like a real message you will send (restock of a specific SKU, not "SALE!!!"). If the mock would embarrass you on a customer phone, rewrite the program before you rewrite the screen.
For broader notification craft beyond the opt-in gate, keep the live guide on ecommerce push notification strategies next to this piece.
Protect the channel after they say yes
Permission is a stock, not a trophy. You can grow it with good asks and burn it with lazy sends.
Segment first, blast almost never
Default to behavior and affinity:
Cart abandoners with a warm cart
Viewed SKU now back in stock
Buyers due for refill based on last order
Category affinity for drops, not the whole file
"Everyone who installed" is a last resort for true brand moments, not Tuesday's promo. DIY builders make the big Send button easy. That is not a strategy. If you need a weekly operator rhythm for what the app should report, pair this with ecommerce app analytics that change weekly decisions.
Frequency caps and quiet hours
Set caps in writing:
Max marketing pushes per user per week
Transactional (shipped, delivered) exempt but still sane
Quiet hours by locale so you are not waking people for a 15% code
When CRM and growth share the channel, one owner enforces the cap. Two teams "just this once" is how opt-out spikes happen.
Topic controls beat all-or-nothing mute
Where the stack allows, let people keep shipping alerts and drop order reminders without eating every campaign. Full opt-out should be easy. Preference center should make full opt-out unnecessary for most.
Deep links that finish the job
A high open rate into a slow webview cart wastes the permission you earned. The tap should land on the cart, PDP, or order status in a real app client. If checkout still feels like mobile Safari inside a shell, fix that path before you scale send volume. See native checkout vs in-app webview.
Weekly metrics that defend the program
Stop leading with open rate alone. Put this set in front of leadership every week.
Opt-in rate
Push-enabled users divided by installers (and separately by app buyers). Track by cohort week so a bad release or a new onboarding flow shows up fast.
Opt-out and permission loss
Weekly opt-outs, OS-level permission off, and uninstall proxies if you have them. A rising open rate with rising opt-out means you are juicing the engaged rump while the base shrinks.
Qualified reach
How many users can receive a marketing push today after caps, quiet hours, and topic prefs. That is your real media inventory.
Push-attributed orders and revenue
Last-tap and assisted views with a consistent window. Split cart recovery, restock, campaign, and transactional-adjacent. If revenue only moves when you attach a heavy discount, you built a coupon pager, not a retention channel.
Reopen after push
D1 and D7 app reopen for users who received a push vs a holdout. You want habit, not only a single redeemed code. Home-screen return is the long game. For why the icon matters beyond notifications, see home screen commerce.
Segment health
Opt-in rate and revenue per send for your top three automated journeys. Kill or rewrite journeys that drive opt-out without orders.
When someone quotes 45% opens or 7x email, attach method: cohort definition, transactional vs marketing mix, and whether the denominator is opted-in users or all installers. Without that, the number is theatre.
What good permission economics look like in practice
You do not need a giant installer base on day one. You need a clean funnel:
Install from owned surfaces (post-purchase, site banner, packaging), not only paid UA.
Soft ask after a clear value moment.
Native dialog with a promise you keep.
Automated journeys that respect caps.
Weekly review of opt-in, opt-out, and attributed orders before anyone adds another campaign slot.
Brands that treat push like a cheaper email blast usually see a short honeymoon, then permission decay. Brands that treat it like a scarce attention channel see the reactivation cost move (in our work, around that -20% band) and keep the list alive long enough for LTV effects to show. App cohorts on ConvertNative programs land near x2.8 LTV ($85 vs about $30 on comparable mobile web) and about +34% conversion when the full native path is in place. Push is one input to that system, not a standalone miracle.
Scale reference: Notino at 7M downloads and €1.2B retail volume. Your file is smaller. The discipline is the same. Permission quality compounds. Blast culture does not.
DIY push consoles vs a managed partner
Most app builders will show you a campaign screen in the demo. Few stay for Monday's opt-out spike, OS permission changes, or the CRM lead who needs segments tied to real purchase data.
A fully managed model (design, store deployment, ongoing optimisation on subscription) matters here because permission work is ops:
Onboarding and pre-permission copy tested after release, not frozen at launch
Journey logic tied to catalog and cart sync
Store and OS changes that affect notification delivery
A partner who will say no to a reckless blast when it protects the channel
ConvertNative production sits around 4–6 weeks when scope is honest. You keep merchandising and CRM intent. We keep the native layer and the hygiene that stops push from rotting. For the ownership framing, see fully managed vs DIY ecommerce app builders and what a mobile app subscription actually includes.
A two-week fix plan if your opt-in is weak
Pull installer→opt-in by week for the last 90 days. Mark any release that changed onboarding.
Turn off first-open native prompts. Ship a soft ask after order success or wishlist save.
Rewrite the pre-permission promise to match journeys you can actually run.
Cap marketing frequency and add quiet hours before the next campaign calendar loads.
Put opt-in, opt-out, qualified reach, and push-attributed orders on the same weekly sheet as email unsubscribes.
Hold out 10% of eligible users from one journey and compare orders and opt-outs. Defend the channel with evidence, not vibes.
If mobile is already most of your traffic and email reach keeps sliding, this is not a copy tweak. It is infrastructure for owned attention. The responsive storefront will not give you a lock-screen relationship. A native app will, if you treat permission like an asset.
Bottom line
Push permission rates ecommerce brands can defend come from timing, an honest value exchange, and send discipline after the Yes. Open rate without opt-in quality is a vanity loop. Segment quality beats blast volume. Quiet hours and caps are product decisions, not legal footnotes.
Build the pool on purpose. Measure it weekly. Spend it carefully. That is how an ecommerce push notifications app becomes a retention lever instead of another unread badge.
Get a read on your push permission funnel
Bring your installer count, current opt-in rate, and top three push journeys. We will tell you whether the leak is timing, promise, product path, or send hygiene, and what a managed native app changes in 4–6 weeks.
Book a free mobile app audit with ConvertNative. Clear scope. No blast-theatre.