My framework for diagnosing growth failures during wartime. What to do when CPA stays bad.

Plus an introduction to wartime growth vs peacetime growth.

We run experiments every single week. Experimentation is our default mode. But sometimes things come up that require deeper exploration.

It usually happens like this: performance has gradually been heading in the wrong direction. First, you rightly ignore it as these things usually work themselves out. Then it starts to become a trend but you still allow…