About $100,000
Spent on these visuals from July 2023 to August 2026. That is the account the swipe file is drawn from. Google never got that kind of budget — do not copy a big Google machine we did not run.
15
Paid ads
What works on Facebook, the structure that produced paid charges, and the ten ways to waste money.
BEFORE YOU SPEND A POUND
- No ads that direct-link our domain without written approval — run paid traffic to your own page or funnel, which then sends people through your link.
- Never bid our brand terms (“movement athlete” and variants) without written approval.
- Optimise for Purchase, never for leads — you are paid on the first paid charge, so that is the only event worth buying.
- The structures below are our own account learnings, shared so you skip the expensive mistakes — adapt them to YOUR funnel.
THE ASSETS THAT GO IN THESE ADS
Everything you need to build a paid ad for TMA lives in these four places. You are not expected to make creative from scratch.
Every ad we have made — 1,105 of them, 448 downloadable as full-size stills — searchable by the offer printed on the artwork. 70% off is the biggest set, ready for the window.
302 Meta creatives with what each one actually did. Ours, and yours to use.
73 creatives from our own paid history — copy the layout, write the words yourself.
324 videos catalogued, grouped by testimonial, UGC and everything else.
Stills, app screens and press — real, cleared and hosted.
312 cleared member quotes, searchable, with the claims you may publish.
What works on Facebook and Instagram
Everything below is from our own spend, not a blog post. Four findings have held every single year:
1
A still image beats a produced video.
In every year of our data, statics with one readable claim out-performed video. If you are choosing where to put an hour, put it in the sentence on the image, not in the edit.
2
The claim has to be readable with the sound off, at thumb speed.
Most people never hear your ad and never stop. One short line doing one job beats a montage with a voiceover.
3
Real people out-pull polish on cold traffic.
A genuine progression clip or a real member beats a studio shoot — but only when the first line is the claim, not a slow build.
4
Optimise for Purchase. Never for leads.
Leads are cheap and they do not pay you. You are paid on the first paid charge, so that is the only event worth buying.
Paid ads — copy the structure that produced paid charges
You are paid on the first paid charge after someone uses your link. A click that never pays is not a win. Every setup below is written so the click lands on one lever and can become a commission.
The badge on the image
The creatives we kept returned. On the swipe file you will see pictures at 2×, 3×, and above 20× ROAS, with the purchase count next to it. Copy those. That is the ROAS that matters.
The account average
A few old videos ate a large share of that $100,000 and did not convert. The average is not the lesson. The structure below, plus the ROAS on each winning image, is what to run.
Where the optimisation actually sat
Meta cold: Advantage+ Sales. Budget and learning sit at campaign level. You do not pick interests. You do not split the budget across ad sets. The ad is what you change.
Meta remarketing: Advantage+ audience is off. You already know who they are. Campaign holds the budget. The ad set is one time window only. The ad does not get its own budget.
Google: we do not optimise at campaign level. We optimise at keyword and search-term level. Performance Max is not the starting structure. Search is.
Meta — cold campaign (Advantage+ on)
One Advantage+ Sales campaign
Objective: Sales. Optimise for Purchase, never for Lead, signup, or landing-page view. You are not paid on a lead. One campaign. Campaign budget. Advantage+ audience on. Advantage+ placements on. Broad — no interest stack of “calisthenics / home workout / fitness”.
One learning unit
Do not build five ad sets. A split budget never finishes learning. One ad set, or let Advantage+ run without manual ad sets. Country: where you can actually talk to the buyer. Exclude people who already bought. Destination URL is locked: your Lever 1 link only.
Same offer, different hook
3 to 6 ads. Same URL. Kill the ad that spends and does not start trials. Do not rebuild the campaign. Use a static with one readable claim, plus one real-person or progression clip. The claim has to be readable with the sound off.
Meta — remarketing (Advantage+ audience off)
These people already saw the product and did not pay. Separate campaigns from cold. Do not let Advantage+ go find a new audience for you here.
One job per campaign
Split by what they did: visited, started the path, reached checkout. Sales objective. Campaign budget. Each campaign excludes purchasers. Do not mix a visit and a checkout in the same campaign.
One window per ad set
7 days, then 8–14, then 15–30. Include only that window. Exclude the hotter window from the cooler one so the same person is not bid on three times. If you only have one list, use checkout abandoners 0–7 days and stop. Do not use one mixed “website visitors 180 days” ad set.
Two ads, one message
One static, one video, same sentence. Visits get Lever 1 (trial). Checkout abandoners get Lever 2 (50% page) and zero trial wording — in the primary text, the headline, and the button. Do not reuse the same static on every window. Change the line when the objection changes.
Google — Search first. Not Performance Max.
We ran Performance Max. It only behaves if brand searches are excluded and a real purchase value is passed in. Without that, PMax buys people searching our name and people shopping for a pull-up bar, and you cannot see which. Start with Search. Work the keyword, not the campaign row.
Problem search
This is the campaign most partners should run. One ad group per intent. Phrase and exact match. Your ad talks about a plan, an assessment, a progression — then your Lever 1 link. Pause the keyword that spends and never pays. Leave the other keywords on.
Review / name search
Only if the click lands on your page or video, and that page uses your partner link. Keywords like “movement athlete review”. Do not bid our name into a blank page. Do not send this traffic to the app store.
Remarketing ladder
Only after you have site traffic. Two campaigns, one ad group each. Conversion goal on the campaign: Purchase only. Rung 1: checkout or cart, last 7 days, Lever 1. Rung 2: those people later, exclude rung 1 and buyers, Lever 2, no trial language. Never one mixed audience. This is not PMax.
What fails — do not copy these
- Optimising Meta for leads or “complete registration”. Cheap emails, almost no paid charges. You do not get commission on a lead.
- Five ad sets at a tiny budget each. Learning never finishes. Nothing is proven, so you keep spending.
- Interest targeting as the strategy. It used to feel precise. It starved delivery. Advantage+ plus the creative beat it.
- Advantage+ audience on remarketing. It leaves the people who already almost bought and goes hunting strangers.
- Performance Max as campaign one. It eats brand queries and equipment queries. You cannot steer it with a negative keyword the way you can on Search.
- Equipment keywords: bars, racks, parallettes, “street workout equipment”, “buy a pull-up bar”. Those people want gear. A subscription will not convert. Add them as negatives on every Search campaign.
- Trial wording on the 50% page, or a 50% creative sent to the trial URL. The page and the ad must be the same offer.
- App Store or Play Store as the click destination. Those purchases do not credit a partner.
- Changing budget, audience, and creative on the same day. Pick one. Then leave it alone for a week.
- One image reused on every remarketing window. The cold hook is not the checkout hook.
What actually produces a paid charge
- Purchase as the only optimisation event. Judge the week on paid charges through your link, not on cheap clicks.
- One cold Advantage+ campaign, enough daily budget to leave learning, ads switched off individually.
- A static with one claim the thumb can read. That format beat produced video across our own years of ads. UGC still works when the first line is the claim, not a montage.
- Remarketing that matches the objection: “you looked” gets the trial; “you got to checkout” gets the 50% page. Buyers excluded.
- Google keywords that mean “I want a training plan”, killed one keyword at a time when they spend and do not pay.
- Your link on the web checkout. Disclose the commission near the ad. No fake deadline.
Keywords — paste these, and the ones to block
No keyword guarantees a commission. These are the queries closest to someone who will pay for a plan. Match type: phrase, then exact once a phrase spends. Send problem keywords to your Lever 1 link. Never send them to Lever 2 unless the ad itself is the 50% offer.
| Use — plan intent | Block — will not pay you |
|---|---|
| calisthenics program | calisthenics equipment |
| calisthenics app | pull up bar |
| bodyweight training plan | dip bars |
| home workout no equipment | gymnastic rings buy |
| beginner calisthenics plan | street workout equipment |
| how to learn a handstand | workout pdf free |
| handstand progression | calisthenics jobs |
| front lever progression | the movement athlete login |
| muscle up training plan | movement athlete coupon code |
| bodyweight workout for beginners | free workout app no signup |
Review keywords — only with your own page in front
movement athlete review · the movement athlete app review · movement athlete vs free youtube · is the movement athlete worth it
The ad must open your review, your email, or your video. The partner link sits in that page. Do not bid these into our homepage with nothing of yours on it — you will pay for a click we already rank for, and you will not be able to see who was yours.