Look at any dataset long enough, and you’ll find patterns. Ice cream sales rise at the same time as drownings. A country’s chocolate consumption seems to track its Nobel prize count. People who drink a glass of red wine each evening live longer than those who don’t. Observational data are full of regularities. But turning those patterns into policies that actually make the world better is surprisingly hard. This chapter explains why our instinct to treat “goes together” as “causes” so often turns well‑meaning policies into failures – and what we can safely do instead.