Everything you learned about causal inference in academia is true. It’s also not enough, and most of us doing applied causal inference experience it. , what’s different is the gravity of the decisions that lean on the analysis: not every decision deserves the same level of evidence. Match your rigour and causal inference to the gravity of the decision, or waste resources. Take product discovery. Before building and shipping, many assumptions need validation at several steps. Aiming to nail each answer with perfect causal inference; for what? Moving up one square on a board of many relevant, even necessary , but on their own insufficient decisions. The risk is already spread, hedged, over many decisions, thanks to a process that values incremental evidence, learning and iterations.…