Start with the mechanic that explains most of the money wasted on martial arts ads, because it is not a secret and it is not a trick. When you build a campaign in Meta's Ads Manager you pick an objective and an optimisation event, and the delivery system then goes looking for the people most likely to perform that event. Meta documents this plainly. It is the whole product.
So the platform is obedient rather than clever. Tell it you want leads and it will find you the people in your town who are most willing to fill in a form. Not the people most likely to get in the car on a wet Tuesday. Not the people most likely to still be training in March. The form-fillers, because that is what you asked for, and it will get very good at finding them.
That is why a school can run an ad account with a falling cost per lead and a mat that looks exactly the same as it did in January. Nothing is broken. The campaign is doing precisely what it was instructed to do.
The only number that pays rent
One of our clients put the useful version of this in a message after a good week:
"Btw the leads this week have been great i have closed 100% of the ones that have walked in the door!"
A school owner, messaging Madscope. Source:
t25.png, reproduced as written. He is identified only by initials in the screenshot, so he is not named here.
Read what he is actually measuring. Not leads, not bookings, not cost per click. The ones that walked in the door. Everything upstream of that sentence is admin.
Amber Beaudin, Chief Instructor at A.J. Bartlinski's Karate Supercenter, describes the same distinction as a before and after. She was getting around five leads a week. She now gets twenty or more, and the number she volunteers next is the one that matters: at minimum ten of those book and attend an introductory lesson. She could have said twenty leads and stopped. She counted attendance instead.
Note that she says at minimum ten, which is a floor rather than an average, and she does not say what becomes of the rest. Most of a lead pile becoming nothing is normal, it is not a scandal, and a supplier who tells you otherwise is selling. What matters is whether the ten who arrive cover what the twenty cost.
Decision one: what you tell the platform to optimise for
This is the setting that moves the most money in the whole account, and almost nobody changes it.
The chain from money to student has four events in it. Somebody sees the ad and gives you their details. Somebody picks a time. Somebody physically arrives. Somebody enrols. You can ask Meta to optimise for the first one, and it is the default most schools end up running, because it is the easiest event to fire and it makes the weekly report look excellent.
It is worth knowing why that first event is so easy to fire. Meta's instant forms can pre-fill a person's name, email and phone number from their profile, so submitting one is barely an action at all. Somebody can hand you their details on the bus, between two other things, without ever deciding to visit your school. That is not a flaw in the form. It is what a form is for, and it is why form volume tells you so little about who will turn up.
Optimise for something further down the chain and two things happen. Your cost per lead goes up, sometimes a lot. And the number of humans on your mat goes up, which is the only movement your landlord recognises. Owners abandon this change in week two because they are watching the first number rather than the second.
If you take one thing from this page: decide which event you are paying for before you look at a single report, and then judge the account on that event only. We wrote out the four ways agencies bill against that same chain in pay per lead versus pay per show.
Decision two: the offer is most of the ad
The offer does more work than the image, the headline and the targeting put together, because it decides who is willing to say yes. It is also the part of the campaign you control completely and your agency does not.
Teena Cirincione, replying publicly in a thread about results, is counting the right thing:
"Totally! In 30 years I have never had 28 paid trials on the mat in week 3 of an 8 week cycle!"
Source:
t25.png, a public Facebook comment, reproduced as written. She names herself in the comment, so she is named here; we have not attached a school to her because the screenshot does not state one.
Paid trials. On the mat. In week three. That is a person who has handed over money and then physically turned up, which is a different species from a free-trial claim, and after thirty years she knows the difference well enough to make it the point of the sentence.
A free trial buys volume and costs you show rate. A paid trial buys commitment and costs you volume. Neither one is correct in the abstract; they are correct relative to what your front desk can handle and what your town will pay. Where exactly to set that price is a bigger question than this page can settle, and it turns on your own show rate rather than on anybody's rule of thumb.
The other thing to know about offers is that they wear out, and owners notice before agencies do. This came in from a school heading into its first summer:
"Of course, I like the process and seeing higher-quality leads coming in for their appointments. I would like to hear your thoughts on a different summer promotion or offer we can do? This is our first summer being open, and I hear it can be difficult to sign up new members."
A school owner, messaging Madscope. Source:
t22.png, reproduced as written. Initials only in the screenshot, so not named.
He is happy with the traffic and asking to change the offer anyway, ahead of the season rather than after it. That is the instinct worth copying.
Decision three: the audience is a capacity question
The temptation is to advertise to everyone who might conceivably train. The better question is which of your classes has seats in it on the days you would like them filled.
An audience decision is really a timetable decision, and the campaign will find the limit faster than you expect. This arrived from an owner running a kids campaign:
"Hey Neil. Class was very packed yesterday. Kids campaign is bringing the heat, we need to pause the campaign until I can make a plan for maybe adding more classes to support this."
A school owner, messaging Madscope. Source:
t30.png, reproduced as written. Initials only, so not named.
Nothing went wrong there. The targeting worked, and then it kept working. We have written at length about which age bracket to point the ads at, and the short version is that the bracket you advertise is the bracket that has to absorb the crush.
Decision four: creative, honestly
Here is where we will disappoint you slightly, because we are not going to give you a list of hooks that work.
We do not publish our clients' creative, we do not have a clean before-and-after test that isolates one image from everything else that changed in the same week, and every "seven ad templates that convert" article you have read was written without one either. Inventing a benchmark is easy. We are not doing it.
What we can tell you is the part the ad is actually being asked to do, and one owner put it better than we would:
"Yes for sure, just dm me. I can say openly in short i was skeptical but went here bc i felt good about risk. Its been an explosion at my gym from 2nd day ads started! And im doing it for my BJJ class which is harder to market than traditional martial arts generally. I regret so bad stalling for so long before I pulled the trigger!"
Jesse Austin, in a public Facebook comment. Source:
t30.png, reproduced as written, typos included. He names himself in the thread, so he is named here; the screenshot does not state a school and we have not attached one.
Read the middle sentence again. His judgement is that his discipline is harder to market than traditional martial arts, and he is probably right: a parent who has decided their child should "do karate" is shopping, whereas an adult who has never heard the phrase Brazilian jiu-jitsu is not in the market at all yet. That difference is a creative brief. One ad needs to be chosen; the other needs to make somebody want a thing they had not considered by Friday.
The practical version for your school: an ad's job is to make the walk-in feel predictable. Who else will be there, what happens in the first ten minutes, whether anyone will be looking at them. The schools that struggle are usually the ones whose ad is a list of virtues, because nobody books a trial to acquire discipline and respect. They book because the Tuesday 5pm beginners class has space and their neighbour's child goes.
Decision five: what happens in the first ten minutes after the form
The best campaign in your county cannot survive a lead sitting unanswered until Thursday. This is the part that most changes the show rate and it is not in the ad account at all, so we gave it its own page rather than a bullet point here.
The four ways schools burn the budget
- Optimising for the cheapest event in the chain. You will get it, cheaply, for ever. See above.
- Advertising into a timetable with no seats. Two of the quotes on this page are owners pausing a working campaign because the room filled before the plan did. Paid traffic pointed at a full class does not queue politely; it evaporates. That failure has a page of its own.
- Judging the account in week one. Early days are unrepresentative in both directions, and we have collected what the first month with an agency actually looks like, including one owner who got no bookings at all in his first week.
- Changing five things at once after a bad week. New offer, new audience, new images, new budget, on the Monday. Now nothing is attributable and the next bad week teaches you nothing either. Change one, give it long enough to produce a countable number of shows, then change the next.
What owners are actually searching for
Measured on 22 September 2026 through the Google Ads search-volume API for the United States in English, covering the twelve months to August 2026, and reproduced as returned:
| Search term | Avg. monthly searches (US) | Highest month in the window |
|---|---|---|
| martial arts marketing | 170 | 480, October 2025 |
| facebook ads for gyms | 90 | 260, September 2025 |
| martial arts marketing agency | 90 | 260, October 2025 |
| martial arts ads | 70 | 210, September 2025 |
| martial arts facebook ads | 50 | 170, October 2025 |
| facebook ads for martial arts schools | 20 | 30, September 2025 |
| martial arts marketing ideas | 20 | 90, September 2025 |
| dojo marketing | 20 | 20, March 2026 |
| how to get more martial arts students | 10 | 10, August 2026 |
Two further phrasings we queried at the same time returned no figure at all from the API and are absent from the table rather than shown as zero.
The honest reading of that table: these are small numbers. Fifty searches a month across the entire United States is a few dozen owners, not a market. What the pattern does suggest is that the looking happens in autumn, and that it happens before the January intake rather than during it. If you intend to run ads in the new year, the account you want in January is the one that has already had a bad week in November and been corrected.
Where we sit in this
We run Facebook and Instagram ads for martial arts schools and we are paid per show, which means the whole of decision one is settled for us in advance. We do not get paid for a cheap lead. We do not get paid for a booking. Somebody has to physically walk into your school. That is the reason this page pushes so hard on one number, and you should weigh it knowing that.
It also means the two failures above that we cannot fix are the ones we care about most: a full timetable and a slow front desk. Neither is in the ad account, and both will stop a working campaign dead.
If you want to go through the five decisions against your actual timetable, book a 15-minute call. Bring the class schedule and the number of spaces you can genuinely take this month. That number decides the rest.
The short version
- Facebook optimises for the event you name. Name the wrong one and it will serve it to you efficiently for ever.
- Count shows, not leads. The owner who closed everyone who walked in the door was counting the right thing.
- The offer decides who says yes. Free buys volume, paid buys commitment, and offers wear out before agencies notice.
- Your audience choice is a capacity choice. The bracket you advertise is the bracket that has to hold the crush.
- Change one thing at a time, and give it long enough to produce a number of shows worth reading.
This page makes two statements about how the platform works: that Meta's delivery system targets the optimisation event the advertiser selects, and that instant forms can pre-fill contact details from a user's profile. Both were checked on 22 September 2026 against Meta's own documentation and independently cross-checked by two other AI models; all three agreed. A third and a fourth platform claim we drafted came back UNCERTAIN on that check and were removed rather than softened. The search volumes were measured on 22 September 2026 through the Google Ads search-volume API for the United States in English, covering the twelve months to August 2026, and are reproduced as returned; Google reports them in rounded bands, groups close variants, and counts searches rather than buyers. Two further terms returned no figure and are absent from the table rather than shown as zero. The written quotes come from messages sent to Madscope or posted publicly by the owners concerned, transcribed from screenshots t22.png, t25.png and t30.png and read directly from the source images before quoting; typos are the owners' own and have been left in. Owners identified only by initials in a screenshot are not named; the two who name themselves in a public comment are named as they wrote it, with no school attached that the source does not state. Amber Beaudin's figures come from a video published on the Madscope YouTube channel and were confirmed against a second, independent speech-to-text pass of the original recording. No industry benchmark, average conversion rate or survey result appears anywhere on this page, because we have not measured one and will not borrow one. Nothing has been rounded, no timeframe has been invented, and no owner's own qualification has been removed.