You are reviewing user engagement on a consumer app and notice that newer signup cohorts behave differently from older ones. A teammate suggests using cohort analysis to decide whether a recent product change improved engagement, while another suggests running an A/B test instead.
What is the difference between cohort analysis and A/B testing? When would you use each, and why is an A/B test usually stronger for making causal product decisions?