Insights
How many ad creatives do you actually need per month?
More than you are running now, and the number scales with spend. A handful of ads a month cannot sustain a $5M+ account, because winners fatigue and you need a steady flow of new ones to replace them.
Matthew Feng · Published 7 August 2026 · 4 min read
There is no magic number, but there is a wrong one
The honest answer is that no fixed count fits every brand. The volume you need is a function of how much you spend, how fast your ads fatigue, and how often a test turns into a winner. What is easy to say is that four polished ads a month is almost always the wrong number for a brand at scale.
Think of it as a supply line, not a deliverable. The account consumes creative and burns it off. Your job is to replace what burns faster than it fatigues. When supply falls below that rate, growth stalls no matter how well the account is set up.
Why fatigue forces constant new volume
Every ad has a shelf life. It works, then the people most likely to respond have already seen it, then frequency climbs and cost per acquisition follows. This is not a setting you can fix. It is how paid social works.
So the account is always losing winners to fatigue. If you are not feeding new concepts in faster than the old ones decay, your blended cost drifts up on its own. The more you spend, the faster this happens, because you reach the responsive audience sooner. That is why the number scales with spend. A $200k a month account burns through creative far quicker than a $30k one.
Creative volume is the ceiling on your growth. The account can only do as much as the supply of fresh ads you feed it.
Why you test many to find the few that carry spend
Most ads you make will not scale. That is normal and it is the entire reason to test at volume. A small share of concepts become the winners that carry the majority of your spend, and you cannot pick them in advance. You find them by putting many distinct angles in front of the algorithm and letting results decide.
This is a numbers game in the literal sense. If only a fraction of concepts win, and you need several live winners at any time to hold performance, then you have to test many times that number every month just to keep pace. Cut the testing volume and you cut your odds of finding the next winner before the current one fatigues.
Volume here means distinct ideas, not exports. Ten colour swaps of the same static count as one concept to the algorithm and they fatigue together. New angles are what matter: a different hook, a different problem, a different proof, a different format.
Why the "four ads a month" retainer is the bottleneck
Most brands at this size pay an agency or freelancer for a fixed monthly output. Four ads. Maybe eight. The work is often beautiful. It is also nowhere near enough to keep a growing account fed.
The economics are the real problem. When you pay per finished ad, cost per creative climbs into the thousands, and that price forces scarcity. You cannot afford to test widely, so you test cautiously, so you learn slowly. Worse, expensive ads become precious. Everyone gets attached to them, and a precious ad is the enemy of a fast-learning account. The retainer caps your volume by design.
How an in-house engine changes the math
The fix is not a bigger retainer for more finished ads. It is a system that lets your own team produce a high volume of on-brand creative at a low cost per unit, guided by people who know what the account needs.
When the concept is briefed properly first, the hook, angle, proof and format written down, production becomes fast and cheap. Your team can make dozens of distinct ads instead of four, and because each one costs a fraction of an agency asset, you can afford to be wrong most of the time. That is the point of testing. This is how brands get to hundreds of on-brand ads a year rather than a precious handful, and it is the model behind results like a 10x return at Copini and $300k to $3m a month at Freya Meds.
A practical way to set your number
Do not chase a universal figure. Reason from your own account instead.
- Count how many live winners you need to hold performance at your current spend.
- Look at how fast those winners fatigue, in weeks, at your frequency.
- Estimate your win rate: out of every batch you test, how many actually scale. Use your own account history if you have it, or start conservative until you have tested enough to know.
- Work backwards. To keep enough winners live as old ones decay, you must test several times more concepts than winners you need.
For most brands at $5M+ that lands well into the dozens of distinct concepts a month, not four. The exact figure moves with your spend and your fatigue curve. The direction never changes: more than you are making now, and it grows as you scale.