Real-time hypothesis testing, premature stopping

Say I want to start offering a discount for shopping in my store. I want to run a test to see if it's a cost-effective idea. I demand an improvement of $d in average sale $s to compensate for the cost of the discount. I start offering the discount randomly to every second customer. Given the average traffic in my store, I determine I should be running the experiment for at least 4 months to determine the true effect equal …

cost customer datascience demand every hypothesis improvement real-time sale shopping store test testing