A cold prospect has to find you, believe you, book something, and then physically turn up at a building they have never been inside, on an evening they had other plans for. Every one of those is a place they fall out. That's what you're paying for when you buy marketing: the cost of dragging a stranger through all four steps.
Someone who trained at your school for eight months and stopped has already done all four. They know where the car park is. They know what the mats smell like. They know your name, they know whether the 6pm class is any good, and they know what it costs. There's no introduction to make and no trust to build from nothing.
That list is sitting in your billing system right now and most owners have never once exported it.
This isn't a clever growth hack and it isn't free money. It's a small, finite, badly-kept asset with a real return and a hard ceiling. This article is about working it properly, and then being honest with yourself about how far it goes.
Start with the export, not with a campaign
Open whatever you bill through. Zen Planner, Kicksite, PushPress, Spark, Mindbody, a spreadsheet, whatever it is. Export every cancelled or expired membership going back three years.
You want five fields and most systems will give you all five:
- Name, and for a child, the paying parent's name. You're writing to the person who signed, not the eight-year-old.
- Start date and cancellation date. These give you tenure, which is the single most useful number on the list.
- Rank or programme level at the point they stopped.
- Contact details, and where the phone number is a mobile. This matters later and it matters legally.
- Cancellation reason, if your system captured one. Most don't. If yours doesn't, add the field today, so the version of you doing this in two years has it.
Then sort by tenure, longest first. Not by cancellation date. Almost everybody sorts by date because the export arrives that way, and it's the wrong order.
Sort by how far they got, not by when they left
The useful question isn't "how long ago did they leave". It's "how much did they invest before they left", because that's what decides whether there's anything unfinished in their head.
Sorted that way the export falls into rough bands, and they don't want the same message.
They got somewhere and stopped. A year or more, a rank they had to work for, and then a cancellation. Something interrupted it. A house move, a job, an injury, a divorce, a kid's football season. These are the strongest names on the list by a distance, and they're the ones most likely to be quietly embarrassed about not having come back, which is exactly why they won't walk in unprompted. Your message has to make returning easy rather than make training sound exciting. They already know it's good.
They got started and drifted. Three to nine months, first or second rank, and the attendance thinned before the billing stopped. The habit never set. This group is real but softer, and the honest framing is a restart rather than a return.
They never got going. A month or two, or a trial that converted and then evaporated. This group has the lowest return rate on your list, and it's also the group your billing export will make look biggest, because trials and short memberships are the bulk of everybody's churn. Don't let the size of it set your expectations for the whole exercise.
Tenure predicts better than recency here, and the reason is worth saying out loud. Somebody who quit last month because the class time stopped working has a problem that hasn't gone anywhere in four weeks. Somebody who quit two years ago at brown belt has had two years for the reason to expire, for the job to change, for the injury to heal, and for the missing thing to start being noticeable.
The window, and the one the law has a view on
There's a natural window and it's wider than most owners assume. Too soon and the reason they left is still true. Too late and they've built a life with a different Tuesday in it.
Roughly, three months to two years after cancelling is where the work is. Under three months you're usually arguing with an unsolved problem. Past two years, some of them have moved away and some of them aren't the same person.
Separately from all that, and worth knowing before you start: US telemarketing rules carry their own definition of how long somebody stays your customer. Under the FCC's rules, an established business relationship for telephone solicitation purposes means a relationship formed by voluntary two-way communication, on the basis of a purchase or transaction within the eighteen months before the call, or an enquiry or application within the three months before it.
That isn't a marketing best practice somebody invented. It's a regulatory definition, and it does one useful thing for you: it puts an outside edge on the list. Beyond eighteen months since their last payment, a former student isn't a former student in any sense the rules recognise. Treat that as the boundary of the exercise, and read the consent section below for what it does and doesn't permit.
What to send
The instinct is to lead with an offer. A free month, half price back, a comeback week. Resist it, at least for the first group.
A discount is the right tool when price was the objection. Price is rarely why somebody with a purple belt stopped coming. Leading with money on that name does two damaging things. It tells them the thing they valued is worth less than they paid for it, and it tells them this is a mailshot rather than someone noticing they were gone.
What works is narrower, and slower to produce. One message, from a human they can picture, about them specifically, asking one question that's easy to answer.
Something with the shape of this:
Hi [name], it's [your name] from [school]. You were [rank] when you stopped training with us back in [month, year], and I was going through the records and realised we never properly heard what happened. No pitch here, I'm just curious. Was it the timing, or something else?
Four things are doing the work there. It's from a named person. It contains a detail only their own school would know. It doesn't ask them to buy, book or commit to anything. And the question has a one-word answer, which is the only kind a busy person replies to.
The offer, if there's one, goes in the second message, after they've replied and told you why. At that point you're not discounting, you're solving the thing they just named. If they say the 6pm stopped working when their shifts changed, the useful answer is "we added an 8pm on Wednesdays", not "here's 50% off".
Send them in small batches, on a weekday evening, and answer every reply yourself the same night. A hundred names sent properly over three weeks beats six hundred sent in an afternoon, because the replies to the six hundred arrive while you're teaching and go cold.
Who you leave alone
This is the part that gets skipped, and it's the part that protects you. Before you send anything, take these off the list by hand. It's an hour with your own memory and your own records, and nobody else can do it for you.
- Anyone who left owing money that was never settled. A friendly reactivation text to somebody expecting a debt collection call goes badly, in a way that ends up on a local Facebook group.
- Anyone who left after a complaint, an injury, or a dispute with you or an instructor. If there was a conversation you wouldn't want quoted back to you, don't reopen it by text.
- Anyone involved in a safeguarding matter of any kind, in any direction. No exceptions and no judgement calls.
- Anyone who asked not to be contacted. Your system may or may not have recorded it. You'll remember some of them yourself. Write those down permanently this time.
- Anyone who left because they joined another school in town. Less a rule than a courtesy, and self-interested too. That owner will hear about it.
Cutting these names barely shrinks the list, and it removes almost all of the downside.
Consent, before you text anybody
The message above is a text message in most owners' heads, because that's what gets read. Texting is also where a small business can do itself real harm without ever meaning to, so here's the plain version for a US school. This is an account of the rules as they stand, not legal advice, and it's worth twenty minutes with your own lawyer before you send at any scale.
Marketing texts sent by automated means need prior express written consent. Under the TCPA and the FCC's rules, initiating a call or text that includes an advertisement or constitutes telemarketing, to a mobile number, using an automatic telephone dialing system or an artificial or prerecorded voice, requires the recipient's prior express written consent. Not implied consent, and not "they were a customer once".
Having been a customer does not replace that consent. The established business relationship described earlier can exempt a live telephone solicitation from the National Do Not Call Registry restrictions. It does not substitute for the written consent standard that applies to autodialed or prerecorded telemarketing calls and texts. Two different rules, and owners routinely assume the first cancels the second.
The exposure is per message. The TCPA's private right of action provides for actual loss or $500 per violation, whichever is greater, and a court may increase that up to three times for a wilful or knowing violation. Multiply $500 by a list of four hundred and you can see why this section exists. Nobody sends four hundred texts intending to break a rule. They send four hundred texts because the software made it one click.
Email sits under a different statute and a lighter one. CAN-SPAM does not require prior opt-in consent for a commercial email. What it requires is that the message isn't deceptive in its subject line or header information, that it carries a clear and conspicuous way to opt out, that it includes your valid physical postal address, and that you honour an opt-out request within ten business days.
The practical read for most schools: your old-student list is usually a stronger email asset than a text asset. Where you do want to text, a message typed by a human to one person is a different animal from a blast sent by a platform. Keep whatever consent records your billing system holds, note which names came with a signed agreement that covered contact, and get your own advice before you push a list into an SMS tool. The tool will happily send. It won't check.
Now do the arithmetic, and be honest about the ceiling
Here's where this stops being a growth strategy and starts being a project with an end.
Count the names that survived the cuts. Say it's 180. Then work out the return rate you actually get. This article isn't going to hand you an industry percentage for it, because nobody's average tells you anything about your list, your town, or how you left things with those people. Run the first batch of fifty, count the replies, count the ones back on the mat sixty days later, and you'll have a real number that belongs to you.
Then multiply it by what one student is actually worth to you over twenty-four months, which is the number that turns "we got nine people back" into a figure you can compare to anything else you might have done with the same fortnight.
Whatever that total comes to, notice the shape of it. It's a one-off. You harvest the list, and then the list is harvested. Next quarter it has regrown by exactly the number of people who quit in the meantime, which isn't a number you want to see growing. A school that can only fill its mats from its own churn is recycling, not growing, and eventually the arithmetic closes in.
So run it, run it properly, and treat what comes back as what it is. The cheapest enrolments available to you, and the last free ones. After that, every extra body on the mat is somebody who has never heard of you, and getting those is a different job with a different cost.
The three-week version
- Export every cancelled membership from the last three years, with tenure, rank and cancellation date. One hour.
- Sort by tenure, longest first, and split into the three groups above.
- Take out the five categories of name you leave alone. By hand, from memory as much as from records.
- Send the first fifty personally, one at a time, evenings, and answer every reply that night. No offer in the first message.
- Count what came back after sixty days and write the number down somewhere you'll find it next year.
Where we sit in this
We're paid when a new student physically walks into your school, so we've no commercial interest in you working your own old list. We'd rather you did it anyway, and did it first, because it's free, and because it's the clearest way to find out what your own follow-up is like when there's nothing to blame it on.
It also tells you something before you spend anything. If you can't convert people who already know and liked your school, a stranger arriving from an advert will find the same gaps in the same places, and they'll find them with none of the goodwill. Where the other five trials went is about that part of the building.
And when the list is worked out, which it will be, the next students have to come from somewhere they've never been. If you want to talk about that part, book a 15-minute call. We'll ask what the reactivation run produced, because it tells us more about your school than anything you could put in a form.
No industry statistic, benchmark, average or survey result appears on this page, and no client of ours is quoted on it. The return rate you should expect is deliberately not stated, because it is not knowable from anybody's data but your own. The consent rules described are US federal rules as they stood in September 2026, checked against the text of 47 C.F.R. § 64.1200 and 47 U.S.C. § 227 and against the Federal Trade Commission's CAN-SPAM compliance guide for business. They are summarised for a school owner, not stated exhaustively, they do not cover state-level restrictions which in several states are stricter, and they are not legal advice. Get your own lawyer's view before sending marketing messages to a list.