Google recently announced a new open-source R package, CausalImpact, for estimating causal effects in time series. With a designed intervention on a time series, CausalImpact is used to predict the counterfactual, i.e., how the response metric would have evolved after the intervention if the intervention had not occurred.