Google AdsTurn each idea into a hypothesis
- Write every idea as a hypothesis with four parts: the change, the audience, the metric it should move, and the reason you expect the move.
- Reject any idea that cannot name a metric. Without one, the test has no result to read at the end.
- Keep one idea to one variable. A new hook and a new offer in the same ad cannot be separated afterwards.
Score and order the list
- Score each hypothesis on three points: how much it could change the result, how cheap it is to make, and how much the team learns if it fails. Write the score next to the item.
Size the budget before it is queued
- Work out the spend each test needs before it can be read, using the test budget sizing guide.
- A test that cannot reach a readable result on the budget you have should be merged with another idea or wait.
Made-up scenario: an account spends 300 a day on one prospecting audience. Testing two hooks side by side gives each arm half that spend. Testing one hook against a control keeps the full budget on one comparison.
Run the queue on a weekly rhythm
- Keep one active test per audience and campaign at a time, unless the test plan says otherwise.
- Review the active test on a fixed day each week, and close it on the stop rule written at the start, not when the result looks good.
- When a test ends, write the outcome, the decision and the next idea in the backlog row. A test stopped early is marked stopped, with the reason, not as won or lost.
What to write in the client report
- Tests run, won, lost, stopped early and still open in the period, with the backlog size at the start and the end.
- The top three ideas in the queue, each with its hypothesis and the metric it should move.
