Ad testing without wasting the budget

Most creative tests fail before they run, because they test two things that are barely different and then get read after nine conversions. Both problems are fixable in an afternoon.

Test hypotheses, not colours

Changing a button from pink to orange is not a test, it is a mood. A test is two different reasons to buy: "it saves time" against "it saves money", "for beginners" against "for people who already tried everything". If the two creatives could not be described in different sentences, you are not testing anything.

A structure that fits a small account

• One campaign, one audience, four to six creatives, each a different angle.
• Enough budget for roughly 50 clicks per creative before you look at anything.
• Three to four days minimum. Weekends behave differently from Tuesdays.
• Kill the bottom two, keep the top one, and build the next four from the angle that won.

When a result is real

A creative with 3 purchases against one with 1 purchase has told you nothing. Read cost per click and hook rate early, they stabilise fast, and only trust cost per purchase once each creative has cleared about 25 conversions. Below that you are reading weather, not climate.

Keep a creative log

One row per creative: the angle in a sentence, the format, the launch date, the result. After three months this file is worth more than any tool, because it is the only record of what your customers actually respond to. Krevya keeps the same log automatically from what you keep and pass.

You are not looking for a good ad. You are looking for the reason a good ad worked, because that reason produces the next ten.