Most published creative testing advice assumes a budget that reaches statistical significance in a week. If yours does not, following it produces a stream of inconclusive tests and a slow loss of faith in testing itself.

Work out your read window first

Before designing a single test, calculate how many conversions per variant you can expect in two weeks. If the answer is under fifty, you are not running a test — you are running a sequential change and calling it a test. That is fine, as long as you say so out loud.

Test concepts, not colours

On a small budget you can afford to learn one thing per fortnight. Spend it on a genuinely different angle — a different objection, a different buyer, a different promise — rather than on a button colour that will never move enough to be readable.

A fortnightly cadence that works

• Week 1: ship two concepts that disagree with each other.
• Week 2: let them run untouched. Resist the urge to intervene.
• Week 3: keep the winner, write down why you think it won, ship the next challenger.
• Every sixth week: revisit the losers. Some of them lost to the format, not the idea.

Twenty-six learnings a year is a serious body of knowledge. It only feels slow next to a calendar you were never able to run.