Creative
How to test ad creative
With targeting largely automated, the ad itself is the biggest lever you have left. Testing it well is mostly discipline: one question at a time, and a rule for the answer decided in advance.
Most creative testing fails for a dull reason. Too many things change at once, the test ends when someone gets impatient, and the winner is whichever ad happened to have a good Tuesday. A little structure fixes that.
Test in this order
- The angle. The reason someone should care. Price, speed, trust, the result, the problem. A different angle can change results several times over.
- The format. Video against image, a customer speaking against a product shot, a before-and-after.
- The opening. The first frame or first line. See our guide to hook rate and hold rate.
- The details. Headline wording, button text, colours. These move results least. Test them last.
Big differences first. Do not polish the headline of an ad built on the wrong angle.
One question per test
Change one thing between the ads in a test. If ad A has a different angle, format and headline from ad B, you will know which won but not why, and the why is what lets you make the next winner.
Decide the rule before you start
Write down three things before launch:
- The measure. Cost per result, where the result is what you actually want. Not click-through rate.
- The spend. How much each ad gets before you judge it. A common rule of thumb is two to three times your target cost per result. An ad that has spent that much with nothing to show has had a fair chance.
- The time. At least several days, so you cover weekdays and a weekend.
Then keep to it. Stopping a test early because one ad is ahead is the most reliable way to pick the wrong winner.
Keep the test fair
- Run the ads at the same time, to the same audience, toward the same goal.
- Keep the test to a handful of ads. Each one needs enough budget to prove itself.
- Remember the delivery system will favour one ad early and give it most of the spend. That is a prediction, not a verdict. Check the others got enough spend to judge.
Small budgets
If you can only afford a few results a week, you cannot run fine-grained tests, and you should not try. Test two very different ads at a time, let them run for two or three weeks, keep the better one, and replace the loser. Slow testing that reaches a real answer beats fast testing that reaches noise.
Write it down
Keep a simple log: what you tested, what you expected, what happened. After ten tests the log is worth more than any single winner, because it shows you what your audience responds to.
Want these numbers working for you?
Tell us what you want to raise, sell or book, and we will show you what we would do first.