Why one winning ad is no longer enough
Ads wear out. After a few weeks the same people have seen yours several times, they stop pausing on it, and each lead costs more. If your whole budget sits on that one video, your sales drop with it.
The platforms push the same way. Meta and TikTok decide who sees each ad. A talking head video does not reach the same people as an animated one. If all your ads look alike, the platform treats them as one ad, and you keep talking to the same audience. Meta itself recommends supplying ads that differ in form, not only in copy.
And nobody knows in advance which one will work. Out of 9 new ads we launched on our own campaigns in November 2026, 3 held up. One script, produced in several visual styles, gave us costs per lead that ranged from one to two times. The only way to find the winner is to test several ads that look different.
Step 1: audit the ads you run today
Open Ads Manager and look at the ads that spent money over the last 30 days. Sort them by form, not by copy. Three questions are enough.
- Who is talking? You, a customer, an actor, a voiceover, nobody.
- What do we see? A face, the product on camera, a screen, animation.
- What tone? Raw, shot on a phone. Or polished, on brand.
If every ad gets the same answers, you have one ad in several copies. That is the most common case. Write down the empty boxes: those are the formats to add.
Step 2: pick 3 to 5 truly different formats
These are the five formats we run. Each one shows one thing well, and each has a limit.
- AI UGC: an AI-generated person talks to the camera like a customer giving a review. Fast to produce and to multiply.
- Direct motion design: a 20 to 30 second animated video that explains the offer. It shows what a camera can't: software, a service, a price, a comparison.
- Raw talking head: you, or someone on your team, talk to your phone for 30 seconds to 2 minutes. No set, almost no editing. It builds the most trust, and it costs nothing.
- Premium motion design: highly polished animation with depth, a moving camera and screens in perspective. It shows the level of the brand and supports a higher price.
- Video with actors: a scene played by actors. It shows a real-life situation. It takes the longest and costs the most.
| Format | What it does well | Its limit | Cost and turnaround |
|---|---|---|---|
| AI UGC | A customer review, multiplied across faces and hooks | Shows when done poorly, hard with a hand-held product | Low, a few days |
| Direct motion design | Explaining an offer, software, a price | Less human warmth than a face | From $447 with us, 3 business days |
| Raw talking head | Trust, proof there is a real person | Needs someone comfortable on camera | Free, one hour |
| Premium motion design | Brand level, a high price point | Takes longer to produce | Higher, one to two weeks |
| Actors | A lived situation, an emotion | Shoot, likeness rights, hard to change | Highest, several weeks |
Three formats are enough to start. Take one with a face, one without, and a third that changes the tone.
Step 3: keep one message, change the form
Diversifying does not mean rewriting everything. If an ad works, its message is good: the promise, the proof, the offer. Keep that script and produce it in each format. The same line said by you on camera, by a voiceover on animation, by a customer in UGC.
That is what makes the test readable. If the script is the same everywhere, the difference in results comes from the form. You then know which format works for your offer.
Change the opening too. Across more than €650,000 of Meta ad spend we analyzed on client accounts, 3 out of 4 people skip the ad before the third second. That is why every ad we deliver comes with 4 hooks: four different openings for the same video. The platform sends budget to the ones that hold attention.
Step 4: launch in small batches, and leave the winner alone
The usual mistake: pausing the winning ad to make room for new ones, or launching everything at once. Either way you lose your baseline.
- Leave the winning ad where it is, with its budget. Don't touch it.
- Put the new ads in a separate ad set, on a small budget.
- Launch 3 or 4 new ads at a time, no more. Beyond that the budget spreads thin and no ad gets enough views to be judged.
- Keep targeting broad. The video finds its audience, not the setting.
- Wait a few days and a few dozen leads before you judge.
Step 5: judge on cost per customer, not on clicks
An ad can get plenty of clicks, cheap leads, and sell nothing. Look all the way down: how many of those leads are real prospects, and how many buy. Our first metric is cost per qualified lead, not cost per lead.
Then the rule is simple. An ad that is expensive or brings poor leads: turn it off. A cheap ad that brings good leads: raise its budget. An ad in between: let it run a little longer.
Then do it again. Every month, a new batch: a format you have not tried yet, or new hooks on the format that won. Diversifying is not a project you finish. It is a habit.
Where to start on a small budget
Start with two formats. The first one is free: film yourself on your phone and say your offer the way you would say it to a customer. The second needs no shoot: a motion design ad that shows your offer clearly and comes in several hooks.
With us, one motion design ad with its 4 hooks is $447, and three ads are $1,087. The first cut arrives in 3 business days. Prices are on the motion design pricing page. Once those two formats are running, add AI UGC, then a more polished format if your brand needs it.