SMS and WhatsApp reach a customer's phone without an app in the way — through MobileConnect, and through Unified Conversations in Advanced Edition and Marketing Cloud Next for exchanges the customer can reply to. The work that decides whether the channel succeeds happens before the first message: consent per channel and per market, and a plan for what happens when somebody answers.
These are genuinely different channels with different economics, different content capability and different rules of engagement. In several of the markets we work in, WhatsApp is the primary messaging channel and SMS is the fallback — which reverses the assumption most UK and US teams start with.
| Consideration | SMS | |
|---|---|---|
| Reach | Universal. Every mobile number can receive it, with no app and no account. | Only customers who use WhatsApp — which in the UAE, and substantially in the UK, Australia and New Zealand, is most of them. |
| Content | Text, a link, limited length. Deliberately plain. | Rich media, buttons, structured replies. Closer to an app experience than to a text message. |
| Starting a conversation | You can send within the constraints of the consent you hold. | Business-initiated messages use approved templates. That approval is a real workflow with a lead time, and it belongs in the plan. |
| Two-way | Keywords and replies, with Unified Conversations in Advanced Edition and Next turning it into a genuine exchange. | Conversational by design. Customers expect a reply, and will be less forgiving than on SMS when they do not get one. |
| Cost shape | Per message, varying substantially by destination market. | Priced differently again, and by conversation rather than by message in the usual models. |
| Use it for | Time-critical, universal reach: delivery, payment failure, appointment, verification. | Richer service and marketing conversations in markets where it is the default channel. |
| The real question | Not which is better. Which one your customers already use, market by market — and the answer will differ across your footprint. | Most organizations operating in several of our six markets end up running both, with the mix set per market rather than globally. |
Capabilities verified against Salesforce documentation Last reviewed 23 September 2026
SMS and WhatsApp implementations rarely stall on the platform. They stall on consent — because obligations differ by market, the consent you hold for email does not extend to a phone, and retrofitting a consent model onto a running program is considerably harder than designing one.
To be direct about the boundary: your legal or privacy team determines what your obligations are in each market. We design and implement a consent model that satisfies what they set, and we make sure the platform honours it consistently. We do not advise on the obligation itself.
Four design questions, all of which have to be answered before the first message.
Someone who subscribed to your newsletter has not consented to a text message. Treating an email opt-in as covering SMS is the most common mistake in this area and the one with the sharpest consequences — a phone is a more intimate channel and customers react accordingly. The consent model needs a separate, evidenced state per channel.
What is required to send a marketing message to a phone number varies across the markets we work in, and a single global rule will be either non-compliant somewhere or needlessly restrictive everywhere. The platform can hold consent state per market; what it cannot do is decide what each market requires. That comes from your legal team, and then we implement it.
Keyword handling — STOP, HELP and their local equivalents — has to be configured, tested and honoured across every business unit and every journey immediately. An opt-out that takes effect at the next audience refresh is an opt-out that has failed, and on this channel that failure is highly visible to the customer.
When someone asks why they received a message, "the flag was set" is not an answer. What was consented to, when, through which mechanism, and under which wording — captured at the point of collection and retrievable afterwards. This is straightforward to build at the start and close to impossible to reconstruct later.
With the consent model designed, the rest is comparatively tractable. The one that surprises teams is the third — a channel customers can reply to needs somebody to reply.
MobileConnect configuration, the numbers or codes you send from, and for WhatsApp the business account and template approval workflow. Administrative rather than difficult, with lead times that belong in the plan rather than discovered inside it.
Configured, tested and verified across every business unit before a single marketing message goes out. This is not a launch task that can slip — it is the mechanism by which a customer withdraws consent, and it has to work on day one.
Unified Conversations in Advanced Edition and Marketing Cloud Next turns one-way promotion into a two-way exchange. That is a genuine capability upgrade and it is also an operating commitment — a channel that invites replies has to answer them.
SMS and WhatsApp activities belong in the journeys you already run, not in a parallel messaging program. A phone buzz is the most intrusive channel you own, so frequency capping has to span every channel rather than each one separately.
SMS and WhatsApp have the least forgiving feedback loop of any channel. An unwanted email is ignored; an unwanted text at 9pm produces an opt-out, and there is no preference center that recovers it.
All four of these are design failures rather than platform failures, and all four are avoidable before launch.
Treating an email opt-in as SMS consent — The most common and the most serious. A customer who subscribed to a newsletter did not agree to a text message, and their reaction reflects that. Beyond the opt-outs, it is precisely the category of decision your privacy team should have been asked about first.
Using the channel for things that do not warrant a buzz — SMS earns its place on urgency: a delivery, a payment failure, an appointment, a verification code. Used for a weekend promotion, it trains customers that your messages are not urgent — and the first thing they do about that is opt out of all of them.
Inviting a reply nobody answers — Unified Conversations makes the channel conversational, and customers use it. Replies arriving in an unmonitored queue is worse than never having invited them, because the customer now has evidence that you do not read what they send.
Frequency capped per channel — Email, push and SMS each capped separately, each behaving correctly. The customer receives all three in a day and experiences a brand that is not paying attention — and the channel they switch off is the intrusive one.
If these have clear answers, an SMS or WhatsApp launch is largely configuration. If more than two do not, the design work is the engagement — and it is considerably cheaper than the remediation that follows launching without it.
The in-app surfaces: push, in-app messages and inbox, and why the constraint there is the app team.
Where SMS and WhatsApp activities belong — inside the journeys you already run.
Unified Conversations, and which editions carry two-way messaging.
Consent mapping when phone and messaging consent arrives from another platform.
The practice page: architecture, edition selection and how an engagement runs.
A worked example of the architecture-first approach on a real client engagement.
You should not assume so, and this is the single most important question on this page. Consent is generally understood per channel: an email subscription is not agreement to receive a text message, and customers treat a phone as considerably more personal than an inbox.
What your obligations actually are in each market is a question for your legal or privacy team, and we would not answer it for you. What we do is build a consent model that holds a separate, evidenced state per channel and per market, and make sure every send honours it.
A WhatsApp Business account connected to the platform, and approved message templates for business-initiated conversations. Template approval is a real workflow with a lead time, which is worth putting into the plan rather than discovering during launch week.
Two-way conversational capability comes through Unified Conversations, available in Marketing Cloud Advanced Edition and Marketing Cloud Next. Before enabling it, settle who answers the replies — that is an operating decision rather than a configuration one.
It depends on your markets, and the answer will differ across them. SMS reaches any mobile number universally. WhatsApp reaches only customers who use it — which in the UAE, and substantially in the UK, Australia and New Zealand, is most of them, and in some markets it is the channel people expect a business to use.
Organizations operating across several of our markets usually run both, with the mix set per market rather than globally. The useful question is not which channel is better but which one your customers are already on, market by market.
That depends entirely on what you have built. Keyword handling — STOP, HELP and local equivalents — should be configured, tested and honoured immediately, across every business unit, before any marketing message goes out.
Beyond keywords, Unified Conversations lets a customer have an actual exchange, and that is an operating commitment rather than a feature. Replies arriving in an unmonitored queue is worse than never inviting them, because the customer now has evidence you do not read what they send.
Frequency capping that spans every channel rather than sitting inside each one. Email, push and SMS capped separately will each behave correctly while the customer receives all three in an afternoon.
The cap belongs above the journey layer, alongside global suppression, so that one change updates everything. SMS should also carry a higher bar for what justifies a message at all — a phone buzz earns its place on urgency, and using it for anything else is how a program loses its opt-ins.
Yes, and they should be. Both are message activities inside Journey Builder, which means channel selection can be made per contact and per moment rather than by running a parallel SMS program.
The design work that matters sits upstream: consent per channel, suppression shared across channels, and a defensible answer to what each channel is for. Adding the activity to a journey is the straightforward part.
Tell us which markets you send to and what consent you currently hold for a phone number in each. That conversation determines the shape of the whole engagement, and it is better had early.
Serving: US | Canada | UK | UAE | Australia | New Zealand
