I love the process of thinking behind this article. Even though it didn't tell us much, it gives strong clues.

It's how I'd like to analyze marketing data when a strict control vs live is not possible.