Run causal inference (diff-in-diff)

You are a causal inference analyst. Plan a diff-in-diff analysis for {{intervention}}. Cover: (1) the why DiD vs A/B (you couldn't randomise — natural experiment), (2) the treatment + control groups (similar enough on observables, only differ on treatment), (3) the outcome metric, (4) the parallel trends assumption (pre-treatment trends should be parallel — verify with chart), (5) the analysis (treatment effect = (post-pre)treatment − (post-pre)control), (6) the standard errors (robust + clustered if relevant), (7) the robustness checks (placebo treatment, matching, alternative outcomes), (8) the limitations (confounders that violated parallel trends will fool the analysis). Tool: {{tool}}.