Beginners ask days. Days don't matter. Sample size and full week-cycles matter.
My rule:
— Minimum 1 full business cycle. Mon-buyers behave nothing like Sun-buyers.
— Stop only after a fixed-horizon sample target you set BEFORE launch.
Last month I called a winner Day 3 at +18%. Felt great. Ran it the full week. Ended at +2% and not significant. The Tuesday spike was a fluke.
Day 3 of any test you are looking at noise wearing a costume.
Go set a sample target now. Don't peek-and-pick. Report back when it bakes.
Split Test Street
@SplitTestStreet
"How long do I run a test?" Wrong question.
This post was published in the Telegram channel Split Test Street. Subscribe here: @SplitTestStreet.