Nice Blog, if anyone wants ai solutions for startup then visit this link:
AI Solution Company

Sometimes the counterfactual are obvious from domain knowledge. For example, think about an action of sending an email and an outcome of click on a link within that email. In this case, the expected outcome of a click if an email would not have been sent is of course zero. Therefore the causal problem is reduced to a predictive problem. Next, a cumulative gain chart can be easily drawn, depicting the tradeoff between total clicks and total targeted population.