Your creative test budget is lying to you i will not promote

Ran the numbers on my creative tests from this year last weekend. 214 variations tested, 11 winners. The thing that stuck with me wasn't the 5% hit rate. It was how many of the 203 losers never had a chance, and not because they were bad. The tests were too small to prove anything either way. Reframe for this post: creative testing is a math problem first and a creative problem second. Here's how I size tests now. Set the kill rule before you launch. Mine is 2.5x target CPA per creative. Target CPA of $40 means each creative gets $100 before I judge it. Five creatives need a $500 test budget. If you've only got $150, you test one or two creatives, not five. Every extra creative crammed into an underfunded test is one you'll end up killing on vibes. One variable per test. New hook, same body, same offer, same landing page. I once tested a new hook plus a new offer plus a new landing page all at once. It won. I had no idea why. Three weeks to unpick it. Boring single-variable tests are the only ones that teach you anything. Read nothing before $75-100 per creative or 3 full days, whichever comes later. Day 1 numbers are mostly noise. I've watched creatives look dead on day 1 and become the account's best seller by day 4. The number of winners killed on day 2 by impatient buyers is honestly depressing. Judge on the deepest event with real volume. Purchases first. Add-to-carts if purchases are thin. Never CTR alone. A 5% CTR with zero add-to-carts means the algorithm found professional clickers, not buyers. My order: cost per purchase, then cost per add-to-cart, then cost per landing page view. CTR is a diagnostic, not a verdict. Kill fast after the threshold, and replace immediately. Past 2.5x CPA with nothing to show, it's dead, no funeral. The part people skip: a testing ad set should always hold 3-5 live creatives. Winners fatigue, usually in 2-4 weeks on broad audiences. If you don't feed new tests in constantly, your account fills up with dying winners and you wonder why performance decays every month. Keep the lab separate from the retirement home. Testing happens in a testing campaign. Winners graduate to the scaling campaign. Never test inside the ad set that's carrying your revenue, because every failed test drags its average down and you'll start killing good tests to protect the number. Rule of thumb: test budget divided by target CPA. If the answer is under 10, you're not testing, you're gambling. Raise the budget or cut creatives until the math works.