Google AdsChange one element
- Pick one element per test: the hook line, the offer line, the proof line or the call to action.
- Write the hypothesis in one sentence, such as: a proof line with a number will lower cost per new customer more than a benefit line.
- Keep the image or video, the audience, the budget and the placement the same for every variant.
Set the sample rule first
- Decide the minimum impressions or purchases per variant before launch.
- Do not call a winner on day one, when spend is small and the numbers swing.
Made-up scenario: two copy variants each get 2,000 impressions on the first day. Variant B has a click-through rate of 2.1 percent and variant A has 1.4 percent. The sample is too small to call, so the test runs until each variant has 15,000 impressions.
Read on the right metric
- Read each element on the metric it is meant to move. A hook is read on click-through, an offer on conversion rate or cost per purchase.
- Check a second metric too. Copy that lifts clicks but lowers order value can be a loss.
- Mark an unclear result as unclear. Do not round it up to a win.
Log and reuse
- Record the element, both variants, the metric, the sample and the decision in the test log.
- Carry the winner into the next test as the new control, and change one more element.
- Keep losing copy in the library with a note on why it lost, so it is not tried again without a reason.
