28. Extracting Value
The Obvious
• Ensure randomized selection
• Create two controls when
possible
• Manage long tests carefully to
preserve valid results
Increasing pressure to get merchandise selection right
Important to challenge how the results will be applied in the future
Results when mailed a catalog. What’s going on here? When they click thru an email recently, they’re less likely to make a purchase in the following 8 weeks??
For example, add an item to back ordered item and get free shipping is just a cool test, but there’s no control. It either works or doesn’t, and there’s nothing special to measure.