An ad test with a real winner

A mid-flight result is wrong far more often than it looks. Acting on it means pausing the better ad about as often as the worse one.

So this check is quiet on most accounts, and that silence is the finding's honesty. When it does speak, the result is real and the change is safe to make.

The check reads the enabled ads of each live ad group over recent weeks, with clicks and conversions, leaving out disapproved ads, which were never in the test.

Checks that weigh money or clicks judge what they find on your own account's numbers, not on one figure applied to every account, so what counts as a problem on a small account is not the same on a large one. Checks on settings read how the account is set up.

When there is too little to go on, the check says so instead of guessing. A finding always comes with the rows behind it, so you can read the evidence before you decide.

1. In Google Ads, open Campaigns, then Ads, for a period with enough data.
2. In one ad group, compare each ad's clicks, conversions and conversion rate.
3. Check that each ad has enough conversions for the comparison to mean something.
4. Run the numbers through a significance test. If the difference could be luck, there is no winner yet.

Pause the losing ad and write a new challenger with a clearly different angle, so the next test asks a real question.

If no test ever gets enough conversions per ad, stop waiting for one. Write ads that differ in their main promise, and judge them on sense rather than statistics.

• Experiment

The checks next to this one in Ads and creative, each with its own page.

• An ad and its page with no word in common
• Display ads holding less than Google can use
• Ads not using the space Google gives them
• Every video the same length, doing the same job
• Ads Google has refused to show
• Free space in the result, unused