Email, SMS or push: which channel for which message
By Abdalla Kamesh · Published October 4, 2026 · 4 min read
Once you have an app, you have three ways to reach a customer, and the temptation is to send everything everywhere. That is how a customer receives the same shipping update as an email, a text and a notification within ninety seconds, and concludes that you cannot organize yourself.
Here is an allocation that uses each channel for what it is actually good at.
The three channels, compared honestly
| SMS | App push | ||
|---|---|---|---|
| Reach | Everyone with an address | Everyone with a number who consented | Only people with the app who granted permission |
| Cost per message | Effectively free | Per message, and it adds up | Free |
| Length | Unlimited | ~160 characters | ~8 words visible |
| Images | Yes | No (MMS aside) | Icon only |
| Speed to attention | Hours | Minutes | Seconds |
| Tolerance for frequency | Moderate | Very low | Low |
| Best at | Detail and persuasion | Urgency | Immediacy and status |
Two rows decide most allocations: reach (email wins) and speed to attention (push wins). SMS is the expensive middle, and its very low frequency tolerance is the thing people under-weight — an unwanted text feels more invasive than an unwanted email by a wide margin.
Message by message
Order confirmation → email. Needs to be findable later, contains a lot of detail, and doubles as a receipt. Push it too if you like, but email is the record.
Shipping and delivery updates → push first, email as the record. This is push’s best use: it arrives in seconds, on the lock screen, and it is the most-tapped notification any store sends. Keep the email as the durable copy with the tracking link. Why these pay for the app.
Abandoned cart → push at one hour, email at 24 hours. Push catches the customer who was interrupted, within the window where they still remember. Email catches everyone without the app, and can carry the product image and reassurance. Do not do both at the same hour. The full sequence.
Back in stock → push. The highest-intent message a store sends, and it is time-sensitive. Email as fallback for people without the app.
Price drop on a wishlisted item → push.
New arrivals / collection launch → email primary, push secondary. Email can show the collection. Push can say it exists. If you do both, put a few hours between them and make the push a different sentence, not the subject line again.
Sale announcement → email + one push. One. Not three over the weekend.
Delivery exception or a real problem → SMS. This is what SMS is for: something has gone wrong, it is time-critical, and you need to reach the customer whether or not they have your app. Worth the cost precisely because you use it rarely.
Failed payment on an order → SMS or push, plus email. Recoverable revenue, and the customer usually does not know.
Review request → push, after delivery plus a day or two. Timed to the delivered order, not the purchase. Timing and handling.
Newsletter / brand content → email only. Never push. Nobody wants a lock-screen notification about your brand values.
Win-back after 60 days inactive → email, then one push. Then stop.
The rule that prevents the triple-send
One event, one primary channel. Everything else is a fallback for people the primary cannot reach.
Practically:
- If the customer has the app and granted permission, push is primary for status and time-sensitive messages, and email carries the durable record.
- If they do not have the app, email is primary.
- SMS is reserved for exceptions and problems, never for marketing.
This requires your systems to know who has the app, which is the one piece of plumbing worth setting up early. Without it you cannot suppress a duplicate, and duplicates are how people unsubscribe from everything.
Frequency, per channel
- Email: a couple of marketing sends a week is normal for most stores. Transactional unlimited.
- Push: two to three marketing notifications a week maximum; transactional unlimited; triggered notifications count at half weight because the customer’s own action caused them. The frequency case.
- SMS: exceptions only. If you are sending marketing texts weekly, you are spending money to annoy people.
And for all three: quiet hours in the customer’s timezone, not yours.
What each channel costs you when you overdo it
Worth being explicit, because the failure modes differ:
Email — an unsubscribe, or a slow slide into the Promotions tab. Recoverable, and re-permissioning is possible.
SMS — an angry STOP, and possibly a complaint. Expensive per message and per mistake.
Push — disabled notifications, or an uninstall. The least recoverable, because there is no re-permission flow: a customer who turns off notifications is not going to Settings to turn them back on.
That asymmetry is the reason push deserves the most conservative frequency of the three, despite being free to send. Free to send is not free to spend.
A starter allocation
If you want one table to implement this week:
- Transactional statuses → push + email
- Cart recovery → push at 1h, email at 24h
- Back in stock / price drop → push
- Campaigns and launches → email primary, one push
- Problems and exceptions → SMS
- Brand content → email only
- Review request → push, after delivery
Then set up the suppression so nobody gets the same thing twice, and read your opt-out rates per channel monthly. Those two habits matter more than any allocation table, including this one.
Related reading