
Retention Marketing
What Is SMS Marketing? From Basics to Advanced Strategies




Written by Darkroom leardership
14 min read
Sep 15, 2026
SMS marketing is the practice of sending promotional and transactional text messages to customers who have given prior express written consent. Brands use it for launches, restocks, order updates, and win-back offers. In the United States, it requires carrier registration through 10DLC and operates under Telephone Consumer Protection Act rules enforced by the FCC.
In this guide, you will see:
Where the revenue actually sits. Klaviyo data shows triggered flows drive almost half of all SMS revenue from a small fraction of sends, making automation the program rather than the add-on. The same concentration shows up in where retention revenue actually comes from.
What has to happen before your first send: 10DLC brand and campaign registration, and the three things that get campaigns rejected.
What the law requires in September 2026, after the one-to-one consent rule was vacated and the revocation rule was rewritten.
Why more than 100 ecommerce brands are currently defending quiet-hours class actions, and what the disputed legal question is.
How to pick a platform, what to measure, and when text message marketing stops being a channel you can run on the side.
How does SMS marketing work?
Think of SMS marketing as a direct line to your customers, but only if they say it's okay. When someone gives you their number (usually at checkout or through a signup form), they're permitting you to reach out.
Their number goes on your subscriber list, and from there, you can send messages using a platform that ensures your texts are delivered through US carriers using a registered phone number.
Even if your message is perfectly written and follows all the rules, it might never reach your customer if your sending number isn’t registered properly. The behind-the-scenes infrastructure determines whether your texts get delivered.
The infrastructure decides whether messages arrive at all. A compliant, well-written text still fails if the sending number was never registered.
The three kinds of message brands send
The categories are not stylistic. They carry different consent requirements and carrier treatment.
Message type | Example | Consent standard |
|---|---|---|
Promotional | Launch announcement, sale, restock alert | Prior express written consent, obtained before the first message |
Transactional | Order confirmation, shipping update, appointment reminder | Consent tied to the transaction, not to marketing |
Conversational | Two-way support, feedback request, reply handling | Initiated by the customer, or by a consented outbound message |
Blurring them is where programs get into trouble. A shipping notification with a discount code appended is a promotional message and needs promotional consent.
What has to happen for an SMS message to actually arrive?
Five systems sit between the send button and the handset. A message has to clear all five, in order, and four of them can fail without telling you.
This is the part of the channel that marketers inherit rather than choose. Someone set it up, it worked, and nobody looked again until delivery dropped and the reporting said everything was fine. Understanding the chain is what lets you tell a creative problem from an infrastructure one.
Link 1: The platform
Your platform does three jobs: it stores subscribers and their consent records, it builds and schedules the message, and it hands that message to an aggregator with direct connections to Verizon, AT&T, and T-Mobile.
That third job is the one nobody evaluates. Two platforms with identical feature lists can deliver differently, because they route through different aggregators and because some pool your sending reputation with other customers on shared infrastructure while others isolate it.
For delivery specifically, ask three questions during the evaluation. Does it report delivery confirmed by the carrier, or only that the message left the building? Can you use your own branded link domain rather than a shared one? Is your sender reputation isolated from other customers on the same route?
Link 2: The number type
Every US program sends from one of three number types. The choice sets your throughput ceiling, your provisioning timeline, and how much filtering scrutiny you attract, and it is expensive to reverse once subscribers have your number saved.
Number type | Provisioning | Throughput | Best for |
|---|---|---|---|
Long code, ten digits | Days after 10DLC registration | Tiered by trust score, roughly 3,000 to 200,000 daily segments | Most DTC brands, and anyone starting out |
Toll-free, 800 or 888 | Days to a few weeks, separate verification path | Higher than an unregistered long code, lower than a short code | Transactional-heavy programmes and national brands with no local identity |
Short code, five or six digits | Weeks, and materially more expensive | Highest available, with the least filtering risk | Very high volume retail, where a queued send costs real revenue |
Choose on three inputs: your peak send volume on your busiest day rather than your average, how soon you need to be live, and whether a local area code matters to how your brand reads.
The mistake worth avoiding is sizing for today. A brand sending 40,000 messages a month is comfortable on a long code until a Black Friday send needs 400,000 in an afternoon.
Link 3: Registration
Think of registration as your way of showing carriers that your business texts are the real deal. There are two main steps here. First, brand registration tells carriers who you are, kind of like showing your ID. Then, campaign registration explains what kinds of messages you plan to send and how your subscribers agreed to get them.
Here’s something many people miss: registration isn’t a one-and-done process. Even after you’re approved, carriers still monitor what you send. If you registered your campaign for order notifications but start slipping in discount codes, carriers will notice and might filter your messages, even if you got the green light at first.
Registration also caps how many message segments you can send each day. If you go over that limit, you won’t get an error message. Instead, the extra messages just get delayed in a queue. That’s why sometimes a flash sale text might show up after the sale is already over.
Link 4: Carrier filtering
Carriers also have their own spam filters, and this is often the most overlooked reason for poor SMS results.
Several things can trigger filtering. These include message content that does not match your registered use case, using public link shorteners like bit.ly, sudden spikes in volume from a number with no history, and higher complaint or opt-out rates. Some regulated categories always get extra attention.
This problem often goes unnoticed because of how it happens. When a message is filtered, there is usually no error. The carrier accepts the message, drops it, and your platform still shows it as sent. You might see a dashboard with 98% delivery, even if a third of one carrier’s subscribers got nothing.
Two ways to spot this issue. First, check delivery and click rates by carrier, since filtering usually affects only one carrier at a time. Second, if you see a sudden drop in click rates but delivery stays the same, consider it a sign of filtering, not a problem with your message.
Link 5: The recipient's device
Even a message the carrier accepts can fail at the handset. The phone may be off or out of coverage, or the number may have been reassigned to someone else.
However, device-level filtering is more common. Both major mobile operating systems offer spam and unknown-sender screening that routes messages to a separate tab most people never open. A subscriber can also block your number directly, which looks identical to a successful delivery from your side.
This is why a delivery receipt is a weaker signal than it appears. It confirms the handset accepted the message. It says nothing about whether the message was sorted, seen, or read.
Before you can send: 10DLC registration and carrier approval
Application-to-person (A2P) 10DLC is how US carriers identify who is sending business text traffic over standard ten-digit numbers. Registration is neither optional nor fast.
Registering the brand and the campaign
Registration happens in two stages. Brand registration establishes who you are. Campaign registration establishes what you will send and how subscribers opted in: Twilio's documentation specifies it must describe how end users opt in, opt out, and get help.
Throughput is assigned from that registration. Twilio documents roughly 3,000 daily SMS segments for sole proprietor brands, roughly 6,000 for low-volume standard, and a tiered allocation for standard brands based on trust score, rising to 200,000 daily segments for Russell 3000 companies.
For a brand running a launch calendar, that number is a planning constraint. A 400,000-message send does not fail loudly. It queues.

What are the rules for marketing texts in the United States right now?
Three things changed between January 2025 and January 2026: Consent, revocation, and quiet hours.
Consent
The FCC once tried to introduce a rule that would have made you get separate consent from customers for every single seller you work with. But in early 2025, the Eleventh Circuit Court struck down that rule. The court said that as long as your consent form is clear and unmistakable, you don't need to get permission from each seller.
After the court’s decision, the FCC officially removed this rule from its regulations in September 2025.
So what does this mean for you? The basic rules remain the same: you still need clear written consent before sending promotional texts, and your disclosure must be obvious when you collect that consent. If you buy or share contact lists, you now have a bit more flexibility, but it is smart to use that freedom carefully and make sure your consent process is crystal clear.
Revocation
Under 47 C.F.R. § 64.1200, a consumer may revoke consent "by using any reasonable method to clearly express a desire not to receive further calls or text messages." The rule treats a specific set of replies as per se reasonable: stop, quit, end, revoke, opt out, cancel, and unsubscribe.
The regulation also sets a hard clock. Revocation requests "must be honored within a reasonable time not to exceed ten business days from receipt."
One part of the rule has not taken effect yet. The FCC's Second Extension Order from 6 January 2026 delayed the requirement that a revocation for one type of message must also apply to unrelated messages. This rule will now take effect on 31 January 2027. It is a good idea to prepare for this change now so you don't have to update your suppression logic later.
Quiet hours
A Florida law firm has recently filed class action lawsuits against over 100 ecommerce brands, all because they sent promotional texts to customers outside the usual 8 am to 9 pm local time window. These lawsuits are based on a specific rule that says you can't send marketing calls or texts to people's homes before 8 am or after 9 pm.
The big question here is whether these quiet hours actually apply to people who have already opted in to receive texts. After all, if someone has given their consent, are those messages still considered 'telephone solicitations'? Defense lawyers say no, and the Ecommerce Innovation Alliance has even asked the FCC to clarify. As of September 2026, we still don't have a final answer.
The operational answer does not depend on how it resolves. Enforce recipient-local-time sending rather than account-time sending, and keep the send window between 8 am and 9 pm. On the accounts Darkroom takes over, the setting usually exists in the platform and has not been switched on.
How do you choose an SMS marketing platform?
Start with category, not feature list. There are three, and they solve genuinely different problems. The wrong choice is expensive to unwind once subscribers have your number saved and your consent records live inside someone else's database.

CPaaS: a communications platform as a service
A CPaaS sells you the pipe. It provides the phone numbers, the carrier connections, and an interface for sending, and leaves everything above that to you. No campaign builder, no segment editor, and no flow canvas unless your team builds one.
Pros
Complete control over routing, message construction, and how subscriber data moves in and out
Carrier-level delivery diagnostics, which is the deepest visibility of the three categories
Priced per message rather than per contact, so cost tracks usage rather than list size
Handles marketing, one-time passcodes, and operational alerts from a single account
Scales to any volume without a plan change
Cons
No marketer-facing interface, so scheduling, segmentation, and flows all have to be built
Needs engineering time to build and continuing engineering time to maintain
Consent capture, storage, and audit trails are your responsibility to design correctly
No native revenue attribution, so lifecycle reporting has to be assembled elsewhere
A marketing team cannot operate it without a developer in the loop
Right for: product teams embedding messaging inside an app, and brands with real engineering capacity who want the pipe and nothing else.
Ecommerce ESP: an email service provider with SMS built in
An email platform that adds text sends to both channels from one customer record. The same profile carries email engagement, purchase history, and SMS consent, and a flow can move between channels without reconciliation.
Pros
One identity graph, so segments, suppression lists and consent all live in one place
Flows branch across channels, so a sequence can send an email and follow with a text on the same logic
Revenue attribution runs against the same order data as email, so the two channels are finally comparable
Usually already in the stack, so no new contract, integration or reporting definitions
One vendor to hold accountable when a send goes wrong
Cons
SMS feature depth trails the specialists, particularly on list growth tooling and conversational handling
Pricing is normally tied to contact count as well as message volume, so a large dormant list costs money
Less control over routing, and thinner carrier-level diagnostics when delivery drops
Concentration risk: one vendor now carries both of your owned channels
SMS can be treated as a feature rather than a product, so it improves on the vendor's roadmap, not yours
Right for: brands already running email there, which is most DTC between $5M and $100M.
SMS-first platforms
These were built for text first, and the difference shows most in getting people onto the list and handling what they send back.
Pros
The deepest list growth tooling of the three, including two-tap sign-up and dedicated capture units
Conversational features are first-class rather than bolted on, so replies can be routed and answered at scale
Compliance tooling designed for text specifically, including consent capture and record management
Support teams who work on this channel and nothing else
Roadmaps where SMS improvements are the product rather than a line item
Cons
A second identity graph to reconcile with your email tool, which is the real hidden cost
Suppression and frequency capping have to be coordinated manually across two systems
Another contract, another integration and another set of reporting definitions
Attribution disputes between the SMS platform and the ESP are common and rarely resolvable
Doubles the surface area for a consent record to go missing
Right for: brands where text is a major revenue line in its own right, and where list growth rather than message quality is the binding constraint.
How do you start a text program, step by step?
Four steps, in order: collect consent properly, segment on purchase behavior, write to the character limit, and set cadence against opt-out rate. The sequence matters because each step constrains the next, and the first one also determines whether your 10DLC campaign gets approved.
1. Build the list before you build the calendar
From what we’ve seen with DTC retention programs, checkout is usually the top performer for collecting new subscribers. Oddly enough, it’s also where brands most often set things up incorrectly when we first start working with them.
Here’s something important to keep in mind: a checked box doesn’t count as real consent. And if you use one checkbox for both email and text message marketing, that won’t pass a 10DLC review.
Separate the two, clearly state what kind of messages people will get and how often, and always include links to your terms and privacy policy right next to the sign-up field.
Every spot where you collect sign-ups needs to have the right disclosure language right next to it. This isn’t just a best practice. Our advice? Write your disclosure once, make sure it’s compliant, and then use that same language everywhere you collect subscribers.
2. Segment on behavior
The segments that move revenue are built from purchase recency, frequency, and value, not from age or region. RFM scoring, short for recency, frequency, and monetary value, gives you the framework, and it is the same one that should be driving your email segments.
3. Write for the 160-character unit
A standard SMS segment is 160 characters. Exceeding it splits the message and multiplies the cost.
Weak: "Hi there! We hope you're having a great week. We wanted to let you know that our new collection just dropped and there are some really exciting pieces. Shop now at the link below!"
Better: "New collection is live. 40 pieces, most under $120. Shop first: [link]. Reply STOP to opt out."
The second version is 94 characters, leads with the news, and carries the required opt-out.
4. Set a cadence you can defend
Do not adopt a fixed sends-per-month number. Set cadence against opt-out rate: if it climbs while click rate holds, you are sending too often to people who still like you. If both fall, the content is the problem. That is the core of any sms marketing strategy that survives the first quarter.
Read also: Email Marketing Benchmarks for Ecommerce That Actually Matters in 2026
How do you measure a text campaign?
Five metrics are ideal for measuring a text campaign: delivery rate, click rate, revenue per recipient, opt-out rate, and incremental contribution.

Click rate and revenue per recipient are from Klaviyo's 2026 SMS benchmarks. The 95% delivery and sub-1% opt-out figures are operating thresholds Darkroom uses on client accounts, and we use them to start an investigation rather than as targets to optimize toward.
Running a holdout so the number means something
Incrementality is the only one of the five that answers whether the program caused the revenue rather than merely accompanying it. The mechanics are simple, and almost nobody runs them, which is why most SMS revenue figures presented to a board are overstated.
Hold back a random 10% of the eligible audience from a flow for a full purchase cycle, not a week. Measure total revenue per eligible subscriber across both groups, not revenue attributed to the message. The difference is what the flow actually produced.
Run it on the abandoned-checkout flow first. That is where attributed revenue most overstates incremental revenue, because a meaningful share of those buyers were returning regardless.
Advanced plays: RCS, conversational flows and orchestration
Rich Communication Services (RCS) is the successor protocol, and it is further along than it was. Apple states that RCS requires iOS 18 and "a cellular plan from a carrier that supports RCS messaging on iPhone," and that "availability varies by region and carrier."
On scale, Juniper Research forecast global active RCS users reaching 1.1 billion in 2024, up from 930 million in 2023. Google said in May 2025 that more than one billion RCS messages were being sent daily in the US, as reported by 9to5Google.
The nuance for brands is that consumer RCS and RCS Business Messaging are not the same thing. Branded business messaging on iPhone is established in Canada, the United Kingdom, Japan, Spain, Germany, and France, while coverage elsewhere remains partial. Treat RCS as a 2027 planning item for US DTC rather than a 2026 migration.
Two-way messaging is the underused play available today. Reply handling turns a broadcast into a service channel, and the replies are the cheapest qualitative research a retention team can get. A hundred people answering "why not this time?" after a lapsed-customer send will teach you more about churn than a quarter of dashboards.
Read also: 7 Proven AI Strategies for Email and SMS Marketing
Who should run SMS marketing?
Text is cheap to start and expensive to run badly. The cost drivers are volume, message type, and carrier fees, and our retention marketing cost breakdown for 2026 covers each.
Keep it in-house when you have a lifecycle owner with capacity, an existing consent trail you trust, and a send calendar under about four campaigns a month. Bring in a partner when the flow library is the gap, when 10DLC registration has already been rejected once, or when quiet-hours exposure makes the compliance question a legal question rather than a marketing one.
Darkroom's retention programs have produced 85% gains in customer lifetime value and 50% year-one revenue growth for DTC brands, with programs live in 30 days. If text is the channel you keep meaning to start, a retention marketing agency can have it live and compliant before the next quarter.

