Meta ads creative testing: how many new creatives a week do you need?

Most e-commerce brands spending €1,000 to €5,000 a week on Meta need roughly 3 to 15 new creatives a week. The right number for you comes out of your budget, whatever the current trend says. Set aside about a quarter of spend for testing, give every new ad around three times your target cost per order to prove itself, and the division tells you how many ads you can test. Launch fewer and the account fatigues. Launch many more and no single ad gets enough spend to be judged.
In short
- Formula: creatives per week = (weekly spend × testing share) ÷ (3 × target cost per order).
- Testing share: 20% to 30% of spend on ads younger than two weeks.
- Mix: about one third new concepts, two thirds variations on proven ads.
- Judge each ad once it has spent about three times your target cost per order.
- Most new ads lose. You need the volume to find the few that win before the old ones tire.
Why does the number of creatives matter more than it used to?
Meta now does most of the targeting itself. With broad audiences, Advantage+ campaigns and automatic placements, the system decides who sees your ad, and the creative is the main signal it uses to decide. A video about back pain in an office chair finds different people than a video about the same chair in a designer living room.
So a creative is no longer only the thing people look at. It is your targeting as well. An account with three ads effectively has three audiences. An account with thirty varied ads gives the system thirty ways to find buyers, and since its 2025 updates Meta's ad delivery has rewarded that kind of variety more and more.
The second reason is fatigue. The people most likely to buy see a winning ad first, and after a few weeks whoever is left responds less. Meta describes it the same way: creative fatigue is when an audience has seen the same creative too many times, which can raise cost per result, and Ads Manager flags an ad as fatigued when its cost per result is at least twice that of your past ads. Cost per order creeps up slowly and most brands blame the auction. In our experience, the cause is nearly always that nothing new went live.
A formula: from testing budget to creatives per week
You can work this out from three numbers you already have, without a benchmark from another brand.
- Weekly ad spend. What you spend on Meta in a normal week.
- Testing share. The part of spend that goes to ads younger than about two weeks. We use 20% to 30%. With less you stop finding new winners, and with more you starve the ads that already sell.
- Test budget per creative. About three times your target cost per order. An ad that spends that much without a single order is very unlikely to become a winner. One that brings in one or two orders at or under target earns more budget.
Creatives per week = (weekly spend × testing share) ÷ (3 × target cost per order).
As rough ranges for a store with a €25 to €40 target cost per order:
- Under €1,000 a week: 2 to 4 new creatives a week.
- €1,000 to €5,000 a week: 3 to 15.
- €5,000 to €20,000 a week: 15 to 40.
- Above that: 40 or more, usually with a dedicated production team.
If you do not have a target cost per order yet, set that first. It is the same number we use to set the ad budget itself.
What counts as a new creative?
A new file is not automatically a new test. We split new creatives into two kinds.
New concepts change what the ad says or who it speaks to. That can mean a different problem, a different buyer or a different format. A founder video, a customer testimonial, a product demo and a comparison static are four concepts.
Iterations keep a proven concept and change one thing, such as the first three seconds, the headline, the thumbnail, the length or the creator. Iterations on a winner have a much higher hit rate than new concepts, but they will not find you a new audience.
A healthy weekly batch is roughly one third new concepts and two thirds iterations on what already works. Make everything new and you rarely build on your winners. Go the other way, with nothing but iterations, and the account slowly narrows to one message and one type of buyer.
Formats matter as well. A mix of short vertical video, UGC-style talking heads, product statics and carousels gives Meta more placements to work with than ten versions of the same static. This is the mix our ad creatives team produces every week.
How do you test without disrupting what works?
Keep the routine simple and stick to it.
- Launch new creatives on a fixed day each week, in one batch, so you compare like with like.
- Put them in a separate testing campaign or ad set. Your proven ads then keep their budget and their delivery history.
- Judge each ad on cost per order once it has spent about three times target. Before that, look only at the early signals that show which ads clearly miss: hook rate (3-second views divided by impressions) and click-through rate.
- Move winners into your main campaign as they are, without rebuilding them from scratch.
- Switch off ads in the main campaign whose cost per order has been above target for a week, once a replacement is live.
Expect most new ads to lose. As an illustration, if one in six new creatives ends up under target, a batch of twelve gives you about two new winners a week, which is enough to keep a mid-sized account fresh.
A worked example
Illustrative figures, not a client result.
Take, as an illustration, a supplement brand spending €4,000 a week on Meta, with a target cost per order of €30.
- Testing budget: €4,000 × 25% = €1,000 a week.
- Test budget per creative: 3 × €30 = €90.
- Creatives per week: €1,000 ÷ €90 ≈ 11.
So this brand launches eleven or twelve new creatives every Tuesday. Four are new concepts (a founder video, a customer story, a morning-routine demo and an ingredients static). The other seven or eight are iterations on last month's two best ads, mostly new hooks and new creators.
With roughly one in six becoming a winner, the brand adds about two proven ads a week, or eight a month, which is enough to replace fatigued ads before cost per order starts to rise. At only three ads a week it would find a new winner every two weeks at best, and the account would slowly get more expensive on the same budget.
Checklist: what a healthy creative testing rhythm looks like
- A target cost per order written down and used to judge every ad.
- 20% to 30% of weekly spend on creatives younger than two weeks.
- A weekly batch on a fixed day, sized with the formula above.
- About one third new concepts, two thirds iterations on winners.
- At least three formats live: vertical video, statics and UGC-style video.
- Each new ad judged after spending about three times the target cost per order.
- Winners moved into the main campaign, and fatigued ads paused only once replacements are live.
- A simple log of every test: concept, hook, format, spend, cost per order.
The batch size also depends on how the rest of the account is set up, and campaign structure and budgets are part of our advertising work. Whatever the setup, launch as many creatives as your testing budget can judge, and do it every single week.
Frequently asked questions
How many ad creatives should I test per week on Meta?
Divide your weekly testing budget by three times your target cost per order. For a store spending €1,000 to €5,000 a week, that usually comes to 3 to 15 new creatives a week.
What share of my Meta budget should go to testing new creatives?
Around 20% to 30% of weekly spend. With less you stop finding new winners, and with more the ads that already sell lose the budget they need.
How long should I let a new ad run before judging it?
Until it has spent about three times your target cost per order. If it has no order by then, switch it off. Before that point, only hook rate and click-through rate tell you anything.
What is creative fatigue?
Creative fatigue is when an ad has been shown so often to the people most likely to respond that results fall and cost per order rises. The fix is new ads, not a new audience.
Is a new hook on the same video a new creative?
It is a variation, not a new concept. Variations on a winner succeed more often, but only new concepts reach new kinds of buyers, so a healthy batch has both.
Do I need video, or are static images enough?
Use both. Statics are quick to make and often win on cost per order, while short vertical video gives Meta more placements and tells more of the story. Accounts with only one format tire sooner.

