Hal Varian on the principles behind causality inference
There are two big problems with the attribution solutions that are around. First of all, they do not provide a baseline (what would happen if you stopped all advertising) and secondly, they measure correlation, not causation.
How To Measure Causal Inference?
Are you keen to get a crash course in how you should structure an experiment to measure causation or causal inference? When you are allowed to apply test result to the entire population? Any marketer that is interested in asking and trying to answer the best questions they could possibly ask and answer should invest 35 minutes in listening to Hal Varians talk.