.ads
← All guides

How many ads should I run?

The variety-vs-concentration tradeoff, and how to test without starving everything.

Enough to give the platform variety, few enough that each ad set gets the data it needs to exit learning. More ads on the same budget means every one of them learns slower.

Two forces pull against each other. Variety helps: the platform's delivery system works best when it has multiple formats and angles to match to different people. But budget concentration also helps: each ad set needs enough results in the learning window to stabilize.

The framework: work out how many ad sets your budget can properly feed (budget ÷ cost per result vs the platform's learning threshold), run that many, and put your creative variety inside them — multiple formats and hooks within an ad set share its learning rather than fragmenting it.

When testing: rotate tests in one at a time against your best performer instead of launching everything at once and starving all of it.

The learning phase is why this matters

When an ad set starts (or gets a significant edit), the delivery system explores: it tries your ad on different people to learn who converts. During that window, performance is noisy and typically worse than it will settle at — judging an ad in its first days is judging the exploration, not the ad.

Two practical consequences: budgets must be able to buy the threshold number of results in the window (see "How big should a budget be?"), and every significant edit — budget jumps, new targeting, swapped creative — resets learning. Frequent tinkering keeps an ad set permanently in its worst-performing state.

How to use it: make changes in batches, give them a week where the platform states one, and resist editing an ad set that is mid-learning unless it is clearly failing.

What we could NOT verify

You'll see confident numbers around the internet — "run 8–15 creatives per ad set". We could not source any such count to official platform guidance, so this product doesn't assert one. Meta's official guidance is qualitative: diversify creative genuinely (distinct visuals, messages, formats) rather than hitting a count. The campaign builder computes how many ad sets your budget can actually feed instead — sign up to open it.

Ranges and trends here are rendered from the same sourced knowledge base the product uses — each with publisher, date and verification stamp inline.