A promotion week often looks worse on short-term ACoS while actually producing a more loyal, higher-LTV customer. You cannot see that in a single week's revenue. Cohort Analysis is the tool that measures the real outcome of a price or coupon test, by following the customers you acquired during the test over the months that follow. This SOP shows you how.
When to use it: whenever you run a price test, or a coupon or Subscribe and Save promotion you want to evaluate properly.
Step 1: Run the test as a clean window
Run the price or coupon change for a defined period so the customers acquired in that window form an identifiable cohort. If you are testing subscriber acquisition, the Overview page chart of Subscribe and Save new-to-brand orders plotted against price and coupon shows the immediate signups for each combination.
Step 2: Open the cohort
Open the LTV & Subscriptions report and go to the Cohort Analysis page.
Find the monthly cohort that matches your test window.
Switch between the views that matter for the test: Units, Profit, Revenue, Retention, etc.
Step 3: Compare against a normal cohort
Compare the test cohort to a cohort from a normal, non-promotion period, reading cumulative performance over the months. A 40% coupon week can look worse on first-order margin but reach higher 6-month profit per customer and better retention. That is the result you are testing for. Use the order type filter to compare Subscribe and Save against non-subscribe cohorts for the same product.
Step 4: Decide on payback, not the week
If the test cohort reaches profitability and retention that beat a normal cohort within a window you are comfortable with, the discount is building value. If it does not, it is just discounting.
Important: cohorts need time to settle. Give the test cohort at least two to three months before drawing a firm conclusion, and avoid judging the most recent weeks, where the second purchase has not had time to happen.