Two channels, different permissions
The difference between SMS and push starts with what each one is addressed to. An SMS goes to a phone number through a mobile carrier. It arrives in the same inbox as messages from friends and family, works on any phone, and needs nothing installed. A push notification goes to an app installation through the operating system's notification service. It appears only if the app is still on the device and the person has allowed notifications, and it disappears into the notification centre if not acted on.
The permissions follow the addresses. Push is governed by the device: the person grants or refuses the notification permission, and can revoke it in settings at any time. Apple also requires that push not be used for promotions unless the person has opted in to that use. SMS is governed by law and by carriers. In the UK, the Information Commissioner's Office direct-marketing guidance explains when electronic marketing messages need prior consent and what an opt-out must look like; in the United States, the CTIA's messaging principles describe what carriers expect of business senders, including clear opt-in and opt-out. Both are general direction, and the rules vary by market, so treat this page as orientation rather than legal advice.
The consequence is that the two channels reach different people. Push reaches the engaged core of app users. SMS can reach customers who never installed the app, people who uninstalled it and people who turned notifications off, provided they consented to SMS. The overlap is smaller than most teams assume, and that is where the choice really starts.
Compare table
SMS vs Push notifications
| Row label | SMS | Push notifications |
|---|---|---|
| Addressed to | A phone number | An app installation |
| Who can receive it | Anyone with a phone who has consented | Installed users who allowed notifications |
| Reach (ReachUnique people who saw an ad.) | Broad across customers, including those without the app | Limited to the engaged app base |
| Permission | Marketing consent under direct-marketing rules and carrier codes | Operating-system opt-in, revocable in settings |
| Cost per message | A per-message charge from the provider, varying by market | Close to nothing beyond the platform licence |
| Format | Short text, sender ID, a link; richer formats depend on carrier support | Title, body, image, action buttons, deep link |
| Where a tap goes | A web page, or the app through a link that opens it if installed | Directly to a screen inside the app |
| Visibility | Sits in the messages inbox alongside personal texts | Banner and notification centre, easily swiped away |
| Measurement | Delivery receipts and link clicks; conversion via tracked links | Delivery, opens, in-app events after the tap |
| Opt-out | Reply STOP or similar; must be honoured promptly | Disable in settings, by topic in-app, or uninstall |
| Tolerance for FrequencyAverage times a person saw an ad. | Low; texts feel personal and intrusive when overused | Moderate; higher for useful, triggered messages |
| Best use | Urgent, transactional, reaching people outside the app | Routine engagement and re-engagement for app users |
The table hides one asymmetry worth stating plainly. A push notification that annoys someone costs a disabled permission. An SMS that annoys someone can cost the number, a complaint to the carrier or regulator, and the channel the business needs for its most urgent messages.
Where SMS wins
SMS wins when the message has to arrive regardless of the app. Delivery updates, appointment reminders, two-factor codes, fraud alerts and payment problems all fall into this group: the person needs the message whether or not they kept the app, and missing it has a real cost. SMS is also the channel that reaches customers who never installed the app, which for many businesses is the majority. A restaurant group, a clinic or a utility usually has far more phone numbers than app users.
It wins for some lapsed users too. Someone who uninstalled the app cannot be reached by push at all; if they consented to SMS, a single well-chosen message with a link that opens the app if it is reinstalled, or the website if not, is one of the few ways back. Treat these as occasional, high-value sends rather than a programme. The guide to re-engaging lapsed app users maps which lapsed users each channel can still reach and measures every return against a holdout.
What SMS does not win is routine marketing. Every message costs money, consent rules are stricter, texts sit beside personal conversations, and tolerance for promotional volume is low. The consent itself has to be built deliberately, with clear wording at the point of collection and records kept, in the same way the list building and consent guide describes for email. Build the SMS programme around a small number of messages people would be annoyed not to receive.
Where push wins
Push wins on cost and depth for anyone who has the app. Because each message costs almost nothing, push can carry the steady flow of activity messages, reminders, back-in-stock alertsA message a shopper asked to receive when an unavailable item, size or colour returns, sent as soon as stock is live and with urgency only where it is true. and personalised prompts that would be unaffordable and intrusive by SMS. Because it opens a specific screen, a tap continues straight into the action the message promised, which is why conversion after a tap (CVRShare of visits or clicks that convert.) tends to be easier to achieve and to measure than for a text link that opens a browser.
It also wins on measurement. Push sits inside the app's analytics, so a team can see not just the tap but what happened next: the screen viewed, the purchase made, the session length, the return a week later. That makes holdout tests straightforward and lets the team judge each message by what it changes rather than by its click-through (CTRShare of impressions that result in a click.). The push notifications and in-app messaging guide covers permission, triggers, caps and testing in detail. The words themselves, a change stated first and a promise the deep link keeps, are the subject of the tutorial on writing push notifications.
Its limits are the mirror of SMS's strengths. Push reaches no one who lacks the app or has turned notifications off, and that group grows over time. A programme that relies only on push slowly loses touch with the customers who drift away, which are often the ones most worth recovering.
Because the push audience is capped by permission, the push permission prompt test is worth running before judging push against SMS on reach, since the push opt-in rateShare of app users shown the notification permission prompt who allowed notifications. it measures sets how many users push can reach at all.
Using both in one lifecycle
The two channels work best as a routing rule rather than two separate programmes. Decide by message type which channel is primary, which is the fallback, and which messages never go by SMS at all. A simple version: push first for everyone who has it enabled; SMS as fallback for urgent and transactional messages where push is not available or has not been acted on within a window that suits the urgency; promotional messages by push and email only, with SMS reserved for a few genuinely valuable occasions for people who explicitly opted in.
| Message | Primary | Fallback | Never |
|---|---|---|---|
| Order or booking confirmation | Push and in-app record | SMS if no push | — |
| Time-critical alert (delivery today, appointment soon) | Push | SMS if no push or no action | — |
| Security codes and account alerts | SMS | — | Marketing content added |
| Activity and back-in-stock | Push | SMS | |
| Re-engagement for app users | Push | Daily SMS | |
| Win-back for uninstalled users | SMS, occasionally, with consent | Push (unreachable) | |
| Promotions | Push, email | SMS for opted-in, high-value occasions | Broadcast SMS to everyone |
Shared rules make the routing work: one frequency cap across all channels, one suppression list, and one record of what each person consented to. The journeys that carry these rules out, including two-way SMS and WhatsApp flows, belong to the automation section; the SMS and WhatsApp automation guide covers that workflow setup. The mobile messaging hub sets both channels beside in-app messages, and the mobile and app marketing guide places messaging in the wider growth loop.
Flickerhallow Cinemas routes ticket messages
Flickerhallow Cinemas, an invented chain, sends a "your film starts in two hours" reminder by SMS to every booking, about 120,000 a month at £0.035 each, or £4,200. Its app accounts for 55% of bookings, and 70% of app bookers have push enabled. Flickerhallow switches app bookers with push to a push reminder, falling back to SMS if push is disabled, and keeps SMS for web and phone bookings. SMS volume drops to about 74,000 a month, costing roughly £2,590. No-show rates stay level at 6% across both routes, and SMS opt-outs fall from 0.3% to 0.1% of recipients because app users no longer receive both channels.
For the retention plan these channels serve, retention through SMS and messaging covers consent, two-way reorders and a frequency budget shared with email.
Common mistakes
- Sending the same message on both channels. Doubles the interruption and the cost. Use a fallback rule instead, so each person gets the message once on the best available channel.
- Treating SMS as cheap push. Per-message cost and stricter consent make SMS the wrong channel for routine engagement. Reserve it for messages people need.
- Relying only on push for urgent messages. People who disabled notifications or uninstalled miss them. Critical alerts need an SMS or email fallback.
- Weak SMS consent. Pre-ticked boxes, vague wording and no record of what was agreed create risk. Collect consent clearly and store it with the date and source.
- Ignoring opt-outs across channels. A STOP reply must be honoured at once, and a person who has opted out of everything should not be found again through another channel.
- Measuring SMS on clicks alone. Many SMS messages work without a click. Measure the outcome they exist for, such as attendance or collection, against a holdout.
- Separate teams, separate caps. If SMS and push are run independently, frequency adds up unseen. Put both under one cap and one suppression list.
