The text went out on Tuesday. By Thursday the replies have slowed to a trickle, a handful of people have written STOP, and the owner is asking the question every owner asks at this point: can we try a different message?
Sometimes that is the right question. It is almost never the right first question. Before anyone rewrites a word, look at who the message actually went to. A text campaign is a message multiplied by a list, and when the result disappoints, the list is where we look first, because it is the half that rarely gets inspected. A better message sent to a dirty list still lands on dead numbers, on patients who already paid, and on people who told you to stop two years ago.
What follows is how we prepare a patient list before a single message goes out, why the order you send in matters as much as the words, the opt-out number your platform is not showing you, and a simple way to tell a list problem from a message problem. There is also a section on when it really is the message, because sometimes it is.
What a raw patient export actually contains
A clinic's patient list is not a list of patients. It is a record of every system the clinic has ever used, every intake form ever filled in, and every person who ever booked, inquired, cancelled or changed their phone number. Export it and you get all of that at once.
You get the same patient twice, once from the old scheduling software and once from the new one. You get phone numbers typed with dashes, with spaces, with a missing digit, and in fields that were meant for something else. You get office lines and home landlines that were never going to receive a text. You get records with a phone number and no first name, which matters the moment your message opens with one. You get the patient who asked to be taken off every list, and the one who booked for next Tuesday, and the one who bought a package last month and is halfway through it.
None of that is anyone's fault. It is what a working clinic's data looks like. But every one of those records either fails silently, reads wrong, or lands an offer on someone who already trusts you and did not need it.
The six removals we make before anyone gets a text
Every list we send for a client goes through the same pass first. It is written into our launch process, it is not optional, and it happens before the list is split or imported anywhere.
- No phone number. Obvious, and easy to miss when the export runs to thousands of rows.
- An invalid phone format. If the number is not a complete, correctly formatted mobile number, the send either fails or goes somewhere you did not intend.
- No first name. A message that opens with the patient's name reads like a mail merge when the name is blank. Fix the record or leave it out.
- Anyone who has opted out or asked to be removed. Every earlier request stands. Messaging them again is both a compliance problem and a trust problem.
- Patients who are already booked. An invitation to book, sent to someone holding an appointment, creates confusion at the front desk and a reply nobody needed.
- Patients who have already prepaid or are partway through a package. This is the one clinics skip, and it is the one that costs the most goodwill. A patient who paid full price last month and then receives an event invitation feels, reasonably, that they paid too much.
Deduplicate while you do it. Then save the excluded records in their own file, with the reason each one was removed. That exclusion file is not bureaucracy. It is how you answer the owner who asks why a particular patient never received the message, and how you avoid re-adding the same bad records on the next campaign.
Only after that pass does anyone start talking about what the message should say. Consent sits underneath all six: the list is people who agreed to hear from your clinic, not people whose number you happen to have. If that distinction is unfamiliar, our piece on what gets a clinic's texting registration denied explains why carriers care about it so much.
Order matters: the warmest patients go first
Even a clean list is not one audience. A current patient who was in the clinic last month, a patient who has not visited in three years, and a lead who inquired once and never came in are three different conversations, and they respond to the same message very differently.
So we do not send to the whole list at once. Current and recent patients go first. Lapsed patients follow. The oldest and coldest records go last, in small batches, if they go at all.
The reason is arithmetic, not courtesy. Messaging platforms judge an account by its opt-out rate across everything it sends. Every STOP from a patient who barely remembers the clinic counts against the same number as a STOP from anyone else. Send the cold segment first and it spends the account's tolerance before the patients most likely to book have even received the invitation. We have also seen opt-outs cluster on the days fresh contacts receive their first message, which is one more reason to control who is in each day's batch.
The same logic applies late in a campaign. When one event sold out, the sold-out message could have gone to the entire campaign list of 2,278 contacts. It went to the 158 people who had actually engaged, in seven batches over about forty minutes, and it drew zero compliance flags. The broad send would have told more people. The narrow one told the right people without putting the account at risk.
The number your platform watches is not the number that matters
Most messaging platforms put one opt-out number in front of you: the share of recipients who replied STOP. It is the compliance number, it is what triggers warnings and restrictions, and it is worth watching closely. It is not the number that tells you how much of your list you just lost.
At one of our early launch events, eight days into the campaign, the STOP rate stood at 1.74%. That was close enough to the platform's roughly 2% warning line that the account's automated sending was temporarily restricted that same day, which is its own lesson. But look at what the STOP figure left out. Another 139 people replied N, our soft decline, on top of the 46 who replied STOP. That is 5.26% more. The real loss to the list was 7%, four times the number on the compliance dashboard.
We changed how we measure after that. Every campaign now tracks two opt-out numbers from day one:
- STOP rate, the compliance number. Our target is 1.5% or lower, because warnings typically start around 2%.
- Combined rate, STOP plus soft declines. Our target is 4% or lower. At 6% we stop and review both the message and which segment it is going to. At 8% the message gets rebuilt.
We also watch the ratio between them. Giving people an easy, visible way to say no, reply N, means most people who are not interested take that route instead of STOP. That early event ran about three soft declines for every STOP. When the ratio falls below one to one, the soft option is not visible enough, and the compliance number will climb faster than it needs to. This connects directly to the thirty day runway before a launch event, because a number that has not been warmed up is far less forgiving of any of this.
How to tell a list problem from a message problem
Once you are tracking the right numbers, the campaign tells you where it is breaking. These are the signals we read, in the order we read them.
- Delivery rate under 95%. List problem. Bad numbers and lines that cannot receive texts. Copy cannot fix a message that never arrived.
- Opt-outs concentrated in one segment. List problem. If the patients who have not visited in years are opting out and recent patients are not, the message is fine and the segment is the issue.
- Opt-outs spiking on the day each new batch gets its first message, across every segment. First-message problem. Look hard at what that message asks of someone who has not heard from you in a while, and make the soft decline easy to see.
- Healthy delivery, low opt-outs, and few replies. Message or offer problem. Our planning target for replies is 10 to 20%, and results vary by clinic and by how recently the list was engaged.
- A STOP that arrives right after a YES. Neither. That is a conversation problem. Someone was interested, asked a question, and got the wrong answer. Those are the most valuable contacts in the campaign and they belong with a person, not another automated reply.
When opt-outs climb, pull the contacts who replied STOP, sort them by time, and look at which message went out just before each one. The pattern is usually obvious within a few minutes, and it points at either a step in the sequence or a slice of the list.
When it really is the message
At that same early event, the list was not the only thing wrong. Opt-outs bunched up on the days new contacts were receiving their first text, which pointed straight at that message. We rewrote it partway through the campaign, and the revised version drew noticeably more clicks and bookings. It is now the starting point for every client, adapted to the clinic and the market.
So yes, rewrite the message when the numbers say to. The point is sequence. Clean the list, order the sends, and measure both opt-out numbers first, and then the message question has an answer you can trust. Rewrite first and you will never know whether the new copy helped or whether it just went to better people. Our post on filling a launch event calendar from your existing patients covers what a good first message looks like.
A patient list is an asset you draw down
The last thing to understand about the list is that it does not refill. Every patient who replies STOP has left that channel unless they choose to come back. Every contact who is messaged too often, or offered the wrong thing, is a little less likely to open the next one. The next event, the next program and the next seasonal offer are all going to be drawn from the same list.
That changes how you make decisions mid-campaign. When a campaign is late, the event is close, and the numbers are turning, the honest question is not how to squeeze out a few more bookings. It is whether continuing is earning bookings or quietly damaging the list for future campaigns. Sometimes the right call is to stop the automated sequence, work the patients who are already engaged by hand, and let the rest of the list rest.
The practical takeaway
Before the next campaign, or before you rewrite the last one, run this pass. Remove the six groups and keep the exclusion file. Split what remains by how recently each patient engaged with the clinic, and send the warmest first. Track STOP and combined opt-outs separately from the first day. Read delivery, segment, first-message and reply signals before you touch the copy. Results vary by clinic, market and database, and no amount of list work rescues an offer nobody wants. But in our experience the list is the cheapest thing to fix and the first place a quiet campaign starts making sense.
Frequently Asked Questions
Why is my clinic's text campaign getting so few replies?
Check the list before the message. Look at the delivery rate first: if fewer than about 95 percent of first messages were delivered, part of the list is bad numbers or lines that cannot receive a text, and no rewrite fixes an undelivered message. Then check whether replies and opt-outs are concentrated in one segment, such as patients who have not visited in years or leads who never came in. If delivery is healthy and the response is weak across every segment, then the message or the offer is the likelier cause. Results vary by clinic, market and database.
What should a clinic remove from its patient list before texting it?
Six groups, before anyone gets a message: records with no phone number, numbers that are not in a valid textable format, records with no first name when the message opens with one, anyone who has previously opted out or asked to be removed, patients who are already booked, and patients who have already prepaid or are partway through a package. Deduplicate while you do it, and save the excluded records with the reason each one was removed, so the decision can be checked later.
What is a good opt-out rate for a clinic text campaign?
Track two numbers, not one. The hard opt-out rate, people replying STOP, is what messaging platforms watch for compliance, and we keep it at or under 1.5 percent because platforms typically begin warning at around 2 percent. The combined rate, STOP plus soft declines such as replying N, is the real loss to your list, and our target for it is 4 percent or less, with a content and segment review at 6 percent. A healthy campaign also sees at least two soft declines for every hard STOP.
Should a clinic text patients who have not visited in years?
Only people who gave your clinic consent to receive texts, and even then, last. Send to current and recent patients first, lapsed patients after that, and older or colder segments in small batches at the end, watching both opt-out numbers as you go. The colder the segment, the more of the account's opt-out tolerance it spends, and that tolerance is shared with the patients most likely to book. Consent rules vary, so confirm your own obligations with qualified counsel.
Want the list cleaned before a single text goes out?
We clean and order the patient list, warm the sending number, write the first message, and watch both opt-out numbers every day of the campaign. See how the system installs, or read the Living Better Healthcare payback case study.
Apply for a Launch Event →