The honest answer: fewer than you think, better than you have
Most brands either run one ad and hope, or upload twenty versions of the same idea and call it testing. Neither works. One ad gives you no comparison. Twenty near copies just split your budget into crumbs and teach you nothing.
Test messages, not decorations
A real test changes what the ad says, not what colour the button is. "Lasts 12 hours" and "made for oily skin" are two different ideas worth testing. The same headline on a blue background and a green background is not a test, it is redecorating. Pick three genuinely different ways of explaining why someone should care, and make one creative for each.
Three to five is a healthy starting point
For most small and mid sized budgets, three to five distinct messages per campaign is enough to learn something real. Give each one enough spend to actually reach people. If your daily budget cannot feed three ads properly, run two and rotate the third in next week. Starving five ads at once is worse than feeding two well.
Let them run before you judge
Killing an ad after one day because it "feels slow" is how good ideas die early. Give each creative a fair window, usually three to five days, before you compare. Look at cost per result, not just clicks. An ad with fewer clicks and more orders is the winner, even if it looks less exciting.
When something wins, make the next round from it
The winning message becomes your new base. Now test variations of that message: a different hook, a demo, a customer question answered up front. This is how creative testing compounds. Every round teaches you something about what your audience actually responds to, and the next brief gets sharper.
The quick checklist
Three to five genuinely different messages. Enough budget per ad to learn. Three to five days before judging. Cost per result over clicks. Winner becomes the base for the next round. That is a testing system, not a guessing game.