How Many Ads Your Meta Account Really Needs

Creative Velocity: How Many Ads Your Meta Account Really Needs

TL;DR: Meta's Andromeda algorithm now picks who sees your ads based on the creative itself, and it groups look-alike ads together. To keep performing, a small business should launch 5 to 10 genuinely different ads every week, test 2 to 3 new ideas weekly, build variations on winners within a week, and refresh top ads every 2 to 3 weeks.

The short answer: more than you think

If you run Meta ads for a small business, you probably need to launch 5 to 10 new ads every week. Not five a month. Not one good video per quarter. Every week.

That sounds like a lot. A few years ago it would have been overkill. But Meta's ad system has changed under the hood, and the businesses that keep winning on Facebook and Instagram are the ones that keep feeding it fresh, varied creative.

This guide explains why, and gives you a simple weekly rhythm you can actually stick to: how many ads to make, how many new ideas to test, and when to double down on what works.

What changed: Andromeda made your creative the targeting

In the old days, you picked an audience (age, interests, lookalikes) and Meta showed your ad to people in that box. Your creative came last.

Andromeda flipped that. It is Meta's ad retrieval system, the layer that scans millions of ads and shortlists roughly a thousand candidates before the final auction decides what someone sees. Instead of starting from your audience settings, it reads the ad itself: the visuals, the words, the audio, the type of person it seems made for. Then it decides who should see it. In plain terms, your creative is now your targeting.

There is a catch. Andromeda groups ads that look or feel alike under a single "Entity ID". Ads in the same group share one ticket into the auction. So 30 ads with the same style, the same setting and the same type of customer count as one ad. Changing the background colour or swapping a headline does not make a new ad in Meta's eyes.

That leads to two rules that drive everything in this guide:

1. Volume matters. More ads means more chances for Meta to find a winner and more fresh options when your current ads tire out.
2. Diversity matters even more. Each ad needs to be genuinely different: a different problem, a different customer, a different emotion, a different format. Otherwise your volume collapses into one ad.

What the data says

The best dataset on this comes from Motion's Creative Benchmarks 2026 report. It covers about 579,000 ads from 6,015 Meta advertisers, launched between September 2025 and January 2026. Three findings matter most for a small business.

1. Winners are rare. Only about 4 to 8% of ads become real winners, meaning ads that take a big share of the budget. For the smallest advertisers (under $10K a month), it is just 3.7%. Roughly half of all ads are switched off within four weeks. That is normal. You should expect most of your ads to lose.

2. Most small advertisers test far too little. Advertisers spending under $10K a month launch about 2.8 new ads per week on average. Those spending $10K to $50K launch about 4.1. Bigger brands launch 11 to 19 a week, and some launch more than 50.

3. The ones who test more find more winners, at every budget level. Within each spend level, the top 25% of accounts launch 1.5 to 3 times more ads than average, and find 2 to 3 times more winners. Mid-sized brands in that top group test about 16 ads a week versus 6.6 on average.

Put those numbers together and the maths is simple. If roughly 1 in 25 ads becomes a winner, then three ads a week gives you about one winner every two months. Eight ads a week gives you one or two winners a month. You are not looking for the perfect ad. You are running enough shots to find it.

Motion is careful to say these are patterns, not proof that more volume causes more winners. But the pattern is consistent across every budget level, and it matches how Andromeda works.

On diversity, the guidance is just as clear. Industry guides now recommend roughly 10 to 20 genuinely different ads active per campaign, and Meta's own testing has reportedly shown one ad set with 25 varied creatives beating five smaller ad sets on both conversions and cost.

Your creative velocity playbook

Here is a rhythm built for a small business spending a few hundred to a few thousand euros a month. It aims at the top 25% of small advertisers, not the average, because average is where most accounts stall.

What is a creative concept? Every new idea you test is a creative concept, built from three choices:

• Persona: who the ad is for (a busy parent, a student, a gym-goer, a retiree)
• Angle: why they should care, meaning the problem, desire or emotion you lead with
• Offer: what they get for acting now (a discount, a free trial, a bundle, a guarantee)

For example: busy parents (persona) who dread deciding what to cook at 6pm (angle), offered 50% off their first box (offer). Change any one of the three and you have a new concept.

Within each concept, make several visually distinct image and video ads that cover the full funnel:

Keep the visuals genuinely different across these ads (different setting, person, format and style) so Meta treats each one as its own ad instead of grouping them.

A useful way to plan: 2 to 3 new concepts a week, each with at least one top, middle and bottom-of-funnel ad, mixing image and video. Add a few variations on your current winners and you land right in the 5-to-10-ads-a-week range.

When to build on winners, and when to test something new

You need both, every week. Testing new ideas finds your next winner. Building on winners squeezes more out of what already works. Do only the first and you never scale. Do only the second and your account slowly runs dry.

A simple split for your weekly batch:

• About half: new ideas. 2 to 3 fresh hypotheses, each in a couple of formats. This is your exploration budget.
• About half: building on winners. New takes on the ideas that are already working.

When to build on a winner. As soon as an ad clearly pulls ahead, usually after about a week and a meaningful amount of spend, start making variations within the next week. Don't wait until it fades. Keep the core idea but change something big enough that Meta sees a new ad: a new hook in the first two seconds, a new format, a different person on camera, or the same message aimed at a different type of customer. Tiny tweaks like a new button colour will just be grouped with the original.

When to start new tests. Every week, without exception. Pick a fixed launch day and treat it like payroll. Give each new ad about a week before you judge it, then switch off the clear losers and let the rest keep running.

When to refresh. Even great ads wear out. Plan for your top ads to lose steam after about two to three weeks, and have their replacements already in testing before that happens.

Once a month, step back. Look at which problems, people, emotions and formats won. Write down what you learned. That becomes the starting point for next month's hypotheses, so your hit rate improves over time.

How to make this much content without a big team

Ten ads a week sounds like a full-time job. It doesn't have to be. The trick is to stop treating each ad as a production and start treating it as an experiment.

• Batch your work. Block two or three hours on the same day every week. Plan the hypotheses, make the ads, launch them. Done.
• Statics count. A strong image with a sharp headline is a real ad, and it takes minutes to make. Not everything needs to be video.
• Use your phone. Lo-fi, phone-shot videos often beat polished ones on Instagram and Facebook because they look like normal posts.
• Mine your customers. Reviews, DMs, support emails and FAQs are full of hooks, pain points and proof. Each one can be a new angle.
• Recycle organic content. Posts that did well on your own page are already tested ideas. Turn them into ads.
• Get real people on camera. Customers, staff or local creators each bring a different face and voice, which Meta reads as genuine diversity.
• Use AI tools. AI ad tools can now produce many on-brand variations in minutes, which makes this weekly rhythm realistic for a team of one.
• Name your files by idea. For example parents-6pm-stress_video_v2. When a winner appears, you'll know exactly which idea to build on.

The bottom line

Meta's algorithm is hungry, and it rewards the advertisers who feed it. Under Andromeda, every genuinely different ad is another way for Meta to find your next customer. Every week without fresh creative is a week your results slowly slip.

You don't need a big budget or an agency to keep up. You need a rhythm:

• Launch 5 to 10 new ads every week
• Test 2 to 3 new ideas every week, each in a couple of formats
• Build 3 to 5 bold variations on every winner within a week
• Keep 10 to 20 genuinely different ads running
• Refresh your top ads every 2 to 3 weeks
• Review what won once a month

Start this week. Pick your launch day, write down three hypotheses, and make your first batch. In a couple of months you'll have something most small businesses don't: a steady supply of winners.