You work on a subscription TV and wireless business and want to know whether a new outbound marketing campaign actually changes customer behavior rather than just reaching people who would have converted anyway. The campaign will be delivered through owned channels such as email and app/web placements, and the team believes it can increase upgrade or add-on purchase conversion among eligible existing customers. You need to design an experiment that can separate true incremental lift from selection bias and short-term noise.
How would you design this experiment so you can credibly estimate whether the campaign improved customer behavior, including how you would randomize exposure, define success, size the test, analyze results, and decide whether to roll the campaign out more broadly?