Somebody showed you a slide with 98% on it. Ninety-eight percent of text messages get opened, the slide said, next to a chart of your email open rate looking sad and gray at 34%. So you turned on SMS. You added it to the same content calendar your emails run on, wrote the first few in the same tone, and pressed send.
Then the bill came.
Not a big bill, necessarily. But a confusing one, because it did not look like anything else you pay for. Your email platform charges a flat monthly fee. Your theme was a one-time purchase. Your ad spend is a number you chose. This one was a number you did not choose, arrived at by a meter you did not know was running.
That meter is the whole story of SMS, and almost nobody explains it before they sell you on the open rate.
Every Other Channel You Own Has a Fixed Price. This One Has a Meter.
Mike Manheimer runs the customer side of Postscript, one of the largest SMS platforms in ecommerce. He is not a critic of the channel. He sells it. And here is how he describes where the cost lands:
"You have all your SaaS apps down here, and then you have your performance marketing spend up here in your P&L, and the SMS bill is in the middle, and it kind of jumps off the page. You're like, wait, this isn't like either of these things really."
He is right, and that sentence is worth more than any open-rate statistic. SMS is not software and it is not media, so it does not behave like either one on your books. Your email tool costs the same whether you send two campaigns this month or twelve. Your text messages do not. The next email you send is free. The next text you send is not.
That single difference should change how you run the channel, and for most stores it changes nothing at all. The texts get scheduled on the email calendar, at email frequency, in email tone, because that is the muscle memory. It is the equivalent of running your paid search budget the way you run your Netflix subscription.
Nobody Has Ever Measured an SMS Open Rate
Start with the number that got you into this.
SMS has no read receipt. The protocol does not have one for marketing messages, and the carriers do not report reads to anyone, because they treat texting as basic infrastructure rather than a marketing channel with engagement reporting. There is no technical mechanism by which anybody could know that 98% of your texts were opened.
So where did the number come from? It is a delivery estimate from the early 2000s, when texting was new and a message on a phone was an event. It got repeated in enough decks that it hardened into a fact. It survives because it sounds great and nobody can disprove it, which is also the reason nobody can prove it.
This is not a reason to skip SMS. A text genuinely does get in front of people in a way an email does not. It is a reason to stop making budget decisions off a number that was never measured. Three things about SMS actually are measurable, and they are the only three that should move your money:
- Clicks. Real, counted, yours.
- Unsubscribes. The most honest feedback any channel gives you, because it costs the customer something to send it.
- Revenue against a holdout. The only one that answers the question you actually have, which we will get to.
You Are Billed in Segments, and You Probably Do Not Know What a Segment Is
Here is the mechanical part that costs stores real money every month, entirely by accident.
Carriers do not bill you per message. They bill you per segment. A segment is 160 characters using GSM-7, the original character set SMS was built on in the 1980s. Go past 160 and your message is split and sent as multiple segments of 153 characters each, and every one of them is billed as a separate send.
Now the part that gets people. If your message contains a single character that GSM-7 cannot represent, the entire message switches to Unicode encoding, and the limit collapses from 160 characters to 70. Concatenated, that is 67 characters per segment.
One emoji does it. So does a curly apostrophe pasted in from a word processor.
So a perfectly ordinary 160-character text with a smiley face at the end is not one segment. It is three. You just tripled the cost of that send across your entire list, and nothing in your dashboard flagged it, because nothing was broken. You were billed correctly for what you sent.
Stack the rest on top: carrier pass-through fees of roughly $0.003 to $0.005 per message that the networks add regardless of your platform, brand vetting around $48 to register, campaign registration in the $15 to $17 range, and monthly carrier fees on top of that. All-in, most US businesses land somewhere between $0.012 and $0.025 per segment.
Run it on a real small store. Two thousand subscribers, eight sends a month, all clean GSM-7 single segments, at two cents. That is $320 a month. Now let a couple of those sends carry an emoji and run long, and the same calendar quietly costs you closer to $500. Same list, same messages, same month.
The Permission Costs More Than the Message
The send is the cheap part. The number is the expensive part.
Marketing texts in the US require prior express written consent, documented and specific to your brand, before you send the first one. Statutory damages under the TCPA run $500 per message, and $1,500 per message where a court finds the violation willful. Per message. That is not a fine for the campaign, it is a fine for each text in it, which is how a single sloppy list import becomes a genuinely business-ending number rather than an unpleasant one.
Most owners have heard some version of this and filed it under "my platform handles compliance." Your platform handles the plumbing. It does not decide where you got the numbers.
There is a softer cost too, and over a year it probably matters more. Somebody unsubscribing from your email list is a shrug. They were on a list, now they are not, and they may well still buy from you. Somebody replying STOP to your text is a door closing, on a phone number you paid to acquire, usually by giving away a discount to get it. You do not get that number back. There is no re-permission campaign for SMS.
A marketer on a panel put the customer's side of it better than any compliance guide:
"You're not my mom, you're not my friend, you're not anyone I know. I use text messaging to communicate with people, not receive ads."
That is your subscriber. She opted in. She meant it at the time. She is one badly judged Tuesday away from being gone permanently.
Everyone Who Tells You How Often to Send Is Selling You the Sends
Ask how many texts a month is right and you will get confident, specific, contradictory answers.
Postscript's position is eight to ten messages per subscriber per month, and they have data behind it: they ran a holdout where half of a list got the normal cadence, a quarter got nothing at all, and a quarter got double, and the double group produced significantly more revenue at 98% statistical certainty. That is a real test, honestly run, and it points at sending more.
It is also a test run by the company that bills you per message. The host of the podcast said so out loud, cheerfully, while praising the result: they were allowed to grade their own homework.
Meanwhile an agency that runs SMS for dozens of eight- and nine-figure brands recommends two campaigns a week and warns that pushing past it drives unsubscribes. And within Postscript's own customer base the spread is enormous. One brand sends 23 messages in a subscriber's first 30 days. Another business at similar scale sends two a month, deliberately, because its owner does not want to pay carrier fees on people who are never going to convert.
All of these people are competent. They are not disagreeing about the facts. They are running different businesses with different margins, different repeat rates, and different things worth interrupting somebody about.
How to Run a Holdout When Your List Is 2,000 People
The holdout is how paid media has always answered this question, and it works at small scale. It is not a big project.
Take 10% of your SMS subscribers at random and exclude them from every campaign send for a full quarter. Not the unengaged ones, not a segment, a genuine random slice. Leave your flows running for everyone so you are testing campaign frequency rather than turning the channel off.
At the end of the quarter, compare total revenue per subscriber between the held-out group and everyone else. Not attributed revenue, which will always flatter the channel. Total revenue from those people, through any door they came in.
If the two numbers are close, your campaigns are mostly harvesting sales that were going to happen anyway, and you are paying per segment for the privilege. If the sending group is meaningfully ahead, you have your answer and you can go find the ceiling by testing more sends against the same holdout.
A quarter feels slow. It is faster than three years of guessing.
Which Messages Are Worth a Segment
Once each send has a price, the question stops being "what should we say this week" and becomes "is this worth interrupting somebody for."
That filter is unusually clean, because the thing SMS is good at is the thing email is bad at: arriving now. The messages that earn a segment are the ones where timing is the entire value.
- The item they wanted is back in stock.
- The order has shipped, or is out for delivery today.
- The sale genuinely ends tonight.
- The drop is live and there are 40 of them.
The messages that do not earn a segment are the ones where nothing would be lost if they arrived on Thursday instead. Brand storytelling. Testimonials. A roundup of your best sellers. The agency above has the receipts on this from its own client data: the nurture-style sends generated the least revenue of any campaign type and pulled the most unsubscribes at the same time. You pay twice, in fees and in list.
The tell is simple. If the message would work as an email, send it as an email. Email is free.
Collect the Number Now, Even If You Never Send a Text This Year
This is the part to act on today regardless of everything above.
Add the phone field to your pop-up and your checkout, with proper consent language, and start collecting numbers whether or not you have any intention of texting anyone this quarter. A list takes months to build and exists the moment you decide you need it. Deciding in November that Black Friday would be a good time to start an SMS program means you have no list in November.
Two things to get right while you build it. Ask for the phone number after the email rather than instead of it, on the second step, where somebody has already committed. And make the ask continue the offer they are already accepting rather than starting a new one. "Finish signing up with text to activate your discount" outperforms "sign up for texts to hear about news," because the first one completes something and the second one starts something.
Keep the consent language accurate to what you will actually send. Consent is specific to your brand and to the kind of messages you described. A list built on a vague opt-in is not an asset, it is a liability with a phone number attached.
This Is Not a Retention Channel. It Is the Second Half of Your Ad Spend.
The last shift is where the money is.
SMS gets filed under retention, run by whoever owns email, and reviewed separately from the ad account. That filing is wrong. Most of the people on your list have never bought anything. They clicked a paid ad, landed on your store, met a pop-up, gave you a phone number, and left without ordering. You already paid for that click.
Which means your SMS program is not a loyalty tool bolted onto the back of the business. It is the second half of an acquisition you have already spent money on, and the only reason it looks cheap is that the ad cost was booked in a different row of the spreadsheet. A text that recovers one of those visitors is not retention revenue. It is the click you bought finally doing its job.
Judge it accordingly. Not against your email open rates, and not against a 98% statistic nobody measured, but against the cost per acquisition you are already tracking on the ad side. That number you have. That number is real. And if you have never worked out what a customer actually costs you once the fees, shipping and returns come out, start there instead, because every decision in this article depends on it.
We run this math on stores all day, and we are happy to run it on yours. If you would rather work it out yourself, the holdout above is the whole method and it costs nothing but a quarter of patience.
