Most analysts start experiment analysis in SQL.

You write a query.
You compare treatment vs control.
You calculate lift.
You declare a winner.

For many well-designed A/B tests, that’s perfectly valid.

Continue reading →