Facebook ads for gyms and fitness studios

Nine image surfaces on this app against Instagram's four, and they do not agree with each other about what an ad should look like. Meta recommends 4:5 at 1440 × 1800 for Facebook Feed, with primary text of 50 to 150 characters and a headline of 27 — and it lists no description line at all. For Facebook Marketplace it recommends a 1:1 image at a minimum of 1080 × 1080, primary text of 125, a headline of 40 and a description of 30. Both are filled from the same ad the moment Advantage+ placements are on. There is no export that satisfies both, which means the choice is which one to lose, and the answer is to write the headline to 27 and compose the 4:5 so that the square crop out of its middle is still the ad.

It is worth knowing how badly a normal run misses that, because it is not close. The ten ads on the Browse page produced fifty headlines between them. Five are inside Facebook Feed's 27 characters; forty-five are inside Instagram's 40. Of the ten headlines the ads actually lead with, exactly one clears 27 — a 22-character line on the win-back ad, which is also the best headline in the batch and almost certainly not a coincidence. Every primary text overruns 150, running from 218 to 542 characters. Nothing is rejected for length; Meta truncates. But truncation happens first on the smallest surface, so the first clause is the ad and everything after it is for the reader who already stopped.

The third difference is shape, and here Facebook is the generous one. It allows 3% aspect-ratio tolerance where Instagram allows 1%, so a crop a few pixels out of true passes here and fails there from an identical file. It also has the surfaces that want a wide asset — the right-hand column and search results — which Instagram does not. Note carefully what that does and does not mean, because this cluster had it wrong until 7 September 2026: Instagram Feed's technical requirements accept aspect ratios from a minimum of 4:5 up to a maximum of 1.91:1, so the wide image is not refused there. What is missing on that app is a placement built to reward it. On Facebook there is, which is why the Browse page carries two 1.91:1 ads, one 1:1 and one 9:16 alongside its six portraits, and its Instagram twin carries only two shapes.

Fourth, the demographic, which is why this page argues about contracts. Facebook reaches 80% of 30- to 49-year-olds against Instagram's 62%, and 74% of 50- to 64-year-olds against 40%. In those bands a gym membership is a recurring household charge that has been signed before, cancelled before, and in a good many cases forgotten about for eight months. That reader is not deciding whether to train. They are deciding whether this particular line on the statement earns its place — which is why the annual fee, the notice period, the pause policy and the off-peak rate are the offers with the most range on this app, and why the one Sales-objective ad in the Instagram batch has no counterpart here. The class pack is the only thing this trade sells that completes at a checkout, and it belongs to the studio.

Finally, the thing that does not change with the app. Fitness is not a special ad category — Meta's Marketing API accepts four, and health and fitness is not among them — so age, radius, lookalikes and exclusions all survive, and every campaign must still declare a value even when that value is NONE. The Health and Wellness standards' 18+ floor attaches to what an ad markets rather than to the trade, exempting fitness services and health clubs marketed for general wellbeing, and the prohibition on statements of inferiority about physical appearance applies everywhere. The Instagram guide is where that argument is made at length, and none of it is weaker here. It is simply not what goes wrong on this app.

Six objectives, and on this app the answer is narrower than on Instagram, for a reason that is really about which half of the trade is here. A boutique studio sells a class pack, which is a fixed-price thing genuinely bought in a browser. A 24/7 access gym sells a membership, which is signed rather than bought — and the access gym is who Facebook reaches. So the distribution tightens toward Leads and away from Sales, and the Browse batch reflects it: seven Leads, two Traffic, one Awareness, and no Sales at all, where the Instagram batch carried one.

That seven-two-one split is the same one the dentist and real estate Facebook batches landed on. Objective is pinned metadata here rather than something the run discovers, so three identical splits are three briefs agreeing rather than an emergent finding — but the agreement is telling anyway. The Facebook cell of every local trade in this cluster is a first-appointment machine, and the appointment is the product. For a gym that appointment is a tour, a trial session or a re-start consultation, and everything below is downstream of getting one booked.

The place to be careful is the second row. Traffic looks like the safe answer on an app full of arguments, and this trade has more arguments than most: the pause policy, the off-peak window, the fee list, the reviews you publish. But Traffic optimises for the click and stops caring, and a gym's whole problem is what happens after the click. Buy it for the two or three claims that genuinely cannot be believed inside an ad, retarget the readers with a Leads campaign, and resist the temptation to make it the default because the landing page looks good in a report.

The order is reversed from the Instagram twin on purpose. That page leads with the studio because its reader is there; this one leads with the 24/7 access gym for the same reason, and the difference is not cosmetic. A studio sells a coach, a capped room and a schedule, and its arguments are proof and story. An access gym sells a building and a contract — price, hours, notice, what happens when you stop coming — and its arguments are objection-handling, mechanism and economics. The six rows below are the access gym's, the next three are the studio's, and the last three fit either. The 'who it is for' line on every row says which.

Each row pins four decisions: who buys it, the angle family that argues it (each one a page in the angle library), the objective to buy it under, and the finished ad on the Browse page that shows it working. Ten rows have an example and two do not. One of the two is the intro offer, which the batch has now skipped twice because it is worthless without a real published price and a real end date. The other is the partner or household membership, which is the offer this app's demographics most obviously support and which nothing in the batch was briefed for — write it first, from your own rate card.

One thing runs across all twelve and it is the same rule the Instagram page ends on: none of them sells a result. The reason on this app is slightly different, though. It is not only that the standards restrict the weight-loss claim; it is that the reader in the 30-to-64 bands has been sold that result before, by a gym they subsequently cancelled, and the argument that survives contact with that memory is an argument about terms.

Start with the declaration, because it is compulsory and it is where an inherited habit costs you the whole account's precision. Meta's Marketing API accepts four special ad categories — housing, employment, financial products and services, and social issues, elections and politics — and fitness is not among them. But every campaign must specify something: pass NONE or an empty array. If you or your agency also runs a housing or a lending account, check the category on the first campaign you build here, because a carried-over HOUSING selection is enforced with a hard error on the audience side and silently costs you custom age, saved audiences, lookalikes and any radius under fifteen miles. A gym giving those away is giving away everything.

The lever that actually decides this app, though, is the one most accounts leave on automatic: placements. Advantage+ placements will render your ad into the feed, Marketplace, search results, the right-hand column, Business Explore, Stories, Reels, in-stream video and video feeds, from one image and one headline, and the recommendations for those surfaces disagree — 4:5 and a 27-character headline in the feed, 1:1 and 40 characters in Marketplace. That is usually the right trade, because restricting placements restricts delivery and a single-site gym has little enough evidence as it is. But it is a trade you should make knowingly: either write and compose to the strictest surface, or turn off the placements you have not designed for and accept the smaller pool. What you should not do is discover in a report that most of your impressions were served somewhere your headline was cut in half.

Then the levers themselves, all of which survive. The thing worth saying about age on this app is that it does real work here in a way it barely does on Instagram: the 30-to-64 bands are where Facebook's advantage is, and they are also where the difference between a 6am lifter, a midday off-peak member and a lapsed member of four years is largest. Band the age to the offer, and set the bottom to your own membership terms rather than to eighteen out of habit. Geography for an access gym is a route rather than a circle — the shift worker trains between work and home, so a pin on the building and a pin on the industrial estate two miles out are both your catchment, and one wide radius is the lazy version of both.

Two cautions carry over from the Instagram guide unchanged, because they are about data rather than about the app. A member list is a usable seed and a usable exclusion, unlike a patient list — but Meta's Business Tools terms prohibit data based on health information including the names given to events, conversions and custom audiences, so name events for the action rather than the goal. 'Trial Booked' is the same event as 'Weight Loss Consult' and only one of them gets flagged. And exclude current members from every acquisition ad set: inside a two-mile radius, an ad offering a held rate to whoever left will reach somebody paying more than that rate on day one, and what arrives is not a lead but a conversation at the front desk.

This app has a destination the other one effectively does not, and for an access gym it is often the right one. Call ads are mobile-only, deliver on Facebook Feed, Marketplace, Reels and Video feed as well as Instagram Feed and Stories, and — this is the part that matters — Ads Manager reports calls placed, 20-second calls and 60-second calls as separate metrics. That last column is the thing no form gives you. A submitted Instant Form tells you somebody tapped; a 60-second call tells you somebody described their week to your front desk, which for a business selling a tour is most of the qualification done before anyone has driven anywhere. If your desk is staffed during the hours you advertise, buy the call and optimise on the 60-second column rather than on volume.

When you buy the form instead, buy the Higher-intent type. Meta's two Instant Form types differ by one step: More volume submits in a thumb movement from pre-filled profile data, Higher intent adds a review screen. That screen is a filter you are paying for on purpose, and this app is where it earns most — a cheap lead from a 55-year-old who half-remembers tapping something is the most expensive kind of lead a gym can buy, because the no-show costs a tour slot and the follow-up call costs the front desk twenty minutes. Add one qualifying question that changes what happens next: which hours you would use, or whether you have trained before.

The landing page is the answer for anything that has to be believed before it can be acted on, and this trade has more of those than most. A pause policy with real terms, a fee list with nothing under it, a published complaints log, an argument about training frequency — none of them survives compression into a form. Send the click, let them read, and retarget with the Leads ad a week later. Class packs and intro offers, where you have them, go to the checkout.

Never the homepage, and on this app there is a second never: not the Facebook Page either. It is tempting, because the Page is where the reviews and the photos and the opening hours already are, and because the ad is already on the same app. But a Page has no offer on it, no form, and a feed of your own posts competing with the one thing you just paid to say. Use the Page as the thing your ad is checked against, which it will be, and send the click somewhere that can convert it.

The first rule is the one this app enforces without telling you: write the headline to 27 characters and compose the portrait so that a square crop out of its middle is still the ad. Meta recommends 27 for Facebook Feed and 40 for Instagram Feed, and 1:1 for Marketplace against 4:5 for the feed, and Advantage+ placements will take one file into both. Of the fifty headlines the Browse batch produced, five fit the Facebook recommendation and forty-five fit the Instagram one — which is what happens when creative made for the other app is posted here, and it is happening in most gym accounts right now.

The second rule is where this batch went wrong, and it is a different failure from the Instagram twin's. Every one of that page's hazards was about the reader's body. Not one of these is. Instead, four of the ten: a membership ad with three invented operating statistics rendered into an email preview, where no alternate headline can reach them; a whiteboard whose caption arithmetic is perfect, whose tally field draws close to a hundred marks under a label reading thirty, and whose body copy quotes an adherence rate against an industry average that does not exist; an equipment ad instructing the reader to scan a QR code drawn by a model that cannot encode one; and an owner's letter describing a competitor's payroll as certainty. The diagnosis is common to all four — the ad reached for proof and invented it. On this app, where the reader is older and comparing terms, the pull toward proof is strong and the cost of faking it is a Page full of recommendations sitting one tap away.

The third is the practical version of the same thing, and it is the rule the previous batch handed over: if a number, a clock, a calendar or a screen carries your claim, composite it from something real. That rule held here in the places it predicted — a fee list whose twelve times $42 is genuinely $504, a training log whose lifts progress consistently across two pages, a review grid where every fix postdates the complaint it answers, and a waiting list that reads identically on a wall board and on a phone in the same frame. It needs one extension. Checking a rendered sum and checking a rendered quantity are two different checks, and the whiteboard passed the first and failed the second in the same image.

Three stages, and on this app the third is bigger than it is on Instagram. Facebook reaches 80% of 30- to 49-year-olds and 74% of 50- to 64-year-olds, which is where the people who used to be members of something are — and the lapsed member is the cheapest acquisition this trade has. They know where the machines are, they left for a reason that may no longer apply, and the objection is embarrassment rather than price. So the shares below shift about ten points out of cold and into the list you already own, which is the opposite of what most single-site gyms do.

Keep the structure this simple deliberately. A single-site budget is small against the evidence Meta's delivery system wants, and every extra ad set divides the same evidence into thinner piles. Variation belongs at the ad level — the access-gym ad and the studio ad live in the same ad set if you run both — not at the ad-set level. And do not build a separate ad set per placement to solve the shape problem; solve it in the creative, which is what the section above is for.

Do the learning-phase arithmetic first, because it tells you how much of Meta's optimisation you are actually buying. An ad set needs roughly 50 optimisation events a week to leave the learning phase; below that it sits in 'learning limited', which is not a penalty so much as the system saying it cannot optimise on what it has. Take your own cost per booked tour, whatever it genuinely is, and multiply by 50: that is the weekly spend at which one ad set has full information. For a single-site gym the honest answer is usually a fraction of it, which is exactly why the funnel above has three ad sets and not nine. Then leave it alone — pausing an ad set, or changing its optimisation event, audience or creative, is a significant edit and restarts learning, and in a trade where the owner is also on the floor, the daily adjustment is the most expensive habit available.

Now the part specific to this app, and it is not the January auction — that argument belongs to the Instagram page, and it is about a studio's peak. The access-gym version of the calendar is a retention problem wearing a seasonal costume. A member acquired in January has a notice period, and if yours is thirty days after a twelve-month term or fourteen days on a rolling contract, the first honest read on that cohort is not available until March or April. Which means a January account that looks superb on cost per join in week three is an account you cannot yet evaluate, and the temptation is to scale it exactly then. Hold the budget flat through the peak, spend the month before it while the auction is quieter and the reader is already deciding, and make the scaling decision when the first cancellation window has closed.

That reframes what to optimise. Cost per join is the number the platform will show you and the one that misleads hardest here, because the cheapest join in this trade is reliably the one that cancels first. The number that decides whether the ads worked is months retained per join, by ad — how many billing cycles the people from each creative actually completed. It takes a spreadsheet and a quarter to see, and it will usually tell you that the ad about the contract outperformed the ad about the discount by a distance that no in-platform metric ever suggested.

One last cost this app has and the other does not. Your ad sits one tap from a Page carrying recommendations, and a business can turn Recommendations off. Turning them off is itself visible, and it is the wrong fix. If you are about to spend meaningfully on Facebook, budget a fortnight of somebody's time to reply to the recommendations you already have before the first ad runs — it is cheaper than the impressions it saves, and the ad on the Browse page built entirely out of answered complaints is the argument for why.

The first two are the ones that cost the most and are almost never noticed, because both fail silently: an ad designed for the wrong frame, and an ad that invented the proof it needed.