There is no magic number of ads under Meta Andromeda. There is a right number for your budget. Here is what Meta has actually recommended, why more ads can hurt as well as help, and a calculator-style method to find your number.
"How many creatives do I need now?" is the question we hear most since Meta Andromeda arrived. Search for an answer and you will find anything from 3 to 50 per ad set to 500 a month. Most of these numbers come from agencies and tool vendors, not from Meta, and many are from accounts spending far more than you.
This article explains what Meta has said, why the right number depends on budget, and a simple method to work out yours.
What Meta has actually said
- Late 2024: for Advantage+ shopping campaigns, Meta recommended at least 20 diversified ads, and putting 20 to 30% of budget towards testing new ads.
- 2025: Meta's guidance shifted to creative diversification as the main lever, and the older rule of thumb of about six ads per ad set no longer appears in its advice.
- December 2025: ads that look or feel alike are treated as variations of the same creative.
- Ads Manager limit: an ad set can hold up to 50 ads. That is a ceiling, not a target.
Notice what is missing: an Andromeda-specific minimum. Meta has said its guidance is about outcomes rather than a particular model. That is why you should treat any fixed number with caution, including the ones below.
Why more ads can help
- More doors. Each genuinely different ad can reach a different pocket of people, so retrieval has more ways to match you.
- Less fatigue. Spend spreads across more ads, so each one is seen less often by the same people.
- More learning. You find out which concepts work for which audiences faster.
Why more ads can hurt
- Spend concentration. Meta usually puts most of the budget on a few ads. Beyond a point, extra ads get almost no delivery and teach you nothing.
- Thin data per ad. With a small budget, twenty ads means each gets too little spend to judge.
- Production quality drops. Teams rushing to hit a number end up making variations, which add little.
- Harder analysis. Too many ads makes it hard to see which concepts are working.
A simple method to find your number
The idea: each concept needs enough spend to show whether it works. So the number of concepts you can run is roughly your budget divided by what each concept needs.
Step 1: Know your target cost per result
Use your target cost per purchase, qualified lead or trial. If you do not know it, work back from customer acquisition cost: what you can afford to pay for a customer, divided by how many results it takes to get one.
Step 2: Decide how much evidence a concept needs
A common rule of thumb is to give a new concept spend equal to two to three times your target cost per result before judging it. Less than that and you are mostly reading noise.
Step 3: Do the maths for a week
Weekly prospecting budget, divided by (target cost per result times 2 to 3), gives the number of concepts you can fairly judge each week.
Illustration. A SaaS company spends 700 a week on prospecting and targets 50 per qualified lead. Each concept needs about 100 to 150 to judge. That supports roughly five to seven concepts at once. The same company at 3,500 a week could fairly run around twenty-five to thirty.
Step 4: Count concepts, then add variations
Your number is concepts, not files. Once a concept proves itself, add two or three hook or format variations of it. Those variations ride on the concept's budget.
Quick quiz
Check your ad count instincts
Four quick questions.
1. Has Meta published a required number of ads for Andromeda?
2. An ad set spends 300 a week and the target cost per lead is 100. Roughly how many concepts can it judge each week?
3. Which adds more reach under Andromeda?
4. What is the maximum number of ads in one Meta ad set?
0 of 4 answered
Recommended ranges by budget
These are starting points from our own practice, not Meta rules. Adjust them with the method above.
- Small (under about 1,500 a month): one prospecting ad set, three to five concepts, refreshed monthly.
- Medium (about 1,500 to 15,000 a month): one or two prospecting ad sets, five to ten concepts each, one to three new concepts every two weeks.
- Large (above about 15,000 a month): consolidated ad sets with ten to twenty concepts, a dedicated testing campaign, and new concepts weekly.
- Advantage+ sales campaigns: follow Meta's late-2024 advice of around twenty diversified ads if your budget can fund them; otherwise start smaller and grow.
How to tell if you have too many or too few
Signs you have too few
- Frequency keeps climbing and results fall within a few weeks of launch.
- One ad takes nearly all the spend and fades quickly.
- Your creative diversity indicator says low.
Signs you have too many
- More than half the ads in the ad set have almost no spend after a week.
- You cannot say which concepts are working.
- New ads never get delivery because older winners hold the budget. Consider a separate testing campaign, covered in Meta ads creative testing.
Frequently asked questions
No. That rule of thumb predates Andromeda and is no longer in Meta's advice. Use your budget to set the number.
If an ad gets almost no delivery after a week, Meta has effectively judged it. Pause it to keep the ad set clean, or move it to a testing campaign if you believe in the concept.
They create variations of your assets. They help optimise a concept but do not replace genuinely different concepts. See creative diversification.
Most accounts do well adding one to three new concepts every one to two weeks, scaled to budget. Small budgets can refresh monthly.
