Every week we audit ad accounts from brands that "tried Meta and it stopped working". The creative library tells the same story almost every time: eight ads, four of them variations of the same concept, all older than two months.
The budget is not the problem. The volume is.
The number one reason brands stop scaling
Meta's algorithm is a signal machine. Every unique creative is a probe into a different corner of your market. When you run five ads, you give the algorithm five probes and it exhausts what it can learn within weeks. Costs creep up, frequency climbs, ROAS sags — and it looks like "Meta got worse" when actually your account went signal-starved.
Across the 30+ Scandinavian B2C accounts we manage, the brands that scale past their plateau share one trait: they never stop feeding the algorithm new creative. The ones that stall almost always test fewer than 15 ads a month.
The formula: angles x concepts
Volume without structure is waste. We build creative volume from two axes:
- Angles — the reasons people buy: pain points, desires, objections, identity. A supplement brand might have 15+ distinct angles (joint pain, vet bills, senior dog guilt, ingredient skepticism...).
- Concepts — the ways you say it: UGC testimonial, street interview, meme static, us-vs-them comparison, founder video, before/after.
Multiply them. 15 angles x 5 concepts = 75 unique ads. That is a month of testing for a brand spending 15,000 EUR. Roughly 20% will beat your account average — those 15 winners are what you scale.
What this looks like in practice
| Monthly spend | Angles to test | Concepts per angle | Ads needed/month |
|---|---|---|---|
| 5,000 EUR | 8 | 3 | ~24 |
| 15,000 EUR | 15 | 5 | ~75 |
| 50,000 EUR+ | 20+ | 6+ | 120+ |
Below these volumes, the algorithm has more budget than signal. Above them, production cost outpaces learning. We tune the ratio per account monthly.
How do you produce that much creative without burning out?
This is the real bottleneck, and it is why we built production formats around volume:
- Statics first. Cheapest probe per angle. A strong static test tells you which angles deserve video budget.
- Street interviews. One shoot day produces 40+ unique video ads — different hooks, different angles, same day. It is the highest ads-per-production-hour format we have found. Full guide here.
- Iteration beats novelty. Half your monthly volume should be children of last month's winners: new hooks on proven bodies, new angles on proven formats.
When we took over Collagen for Hund, the previous setup tested a handful of ads monthly. We moved them to 10+ per week. Sales grew ~10x year over year at ~40% lower CPA — not because any single ad was magic, but because the system never stopped learning.
FAQ
Does testing more ads increase my budget requirements?
No. Testing budget is carved out of existing spend (we typically run test campaigns at 10-20% of account budget). Volume changes how the budget learns, not how much you spend.
Can't I just let Advantage+ figure it out?
Advantage+ optimizes delivery, not creative supply. It can only pick winners from what you feed it. Feeding it three ads and expecting scale is asking it to win a card game with three cards.
How fast should I kill an underperforming ad?
We give statics 3-5 days or ~2,000 impressions in a test cell before cutting. Video gets slightly longer for watch-time signal to accumulate. Winners graduate; everything else dies without mercy.