Setting up a clean AdSense experiment

Setting up a clean AdSense experiment
Use the built-in Experiments tab, but set it up right:
— Change ONE variable per experiment (density OR style, never both)
— Let it run to statistical significance — Google tells you, don't stop early
— Watch the 'Estimated change' but verify against your own RPM in reports
— Apply the winner, then immediately start the next experiment
— Keep a log: date, variable, result. Memory lies after 3 tests
I ran 'ad load' experiments back to back for a month, compounded +30% RPM.
Steal it.