Can you explain a little more? Do you already have data on a past similar campaign to build your model?
Creating the dependent variable depends on the business. Customers who shopped during the campaign period can be defined as the responders. Or customers who bought only the featured items/products during the campaign period can be the responders. The easiest one is where coupons were mailed - in that case, customers who redeemed will be the responders.
Control and Test: To measure campaign performance, similar or comparable customers are randomly assigned into 2 groups. One is mailed or targeted, while the other group doesn't get anything.