TinkerShoppers queue at the turnstile.
Chapter 1 of 7
Did it work?
Larkspur’s team built a one-page checkout. The owner wants proof it sells more before every shopper gets it.
This chapter: How do you tell whether a change worked, and not something else?
About this chapter
You will learn
- Split shoppers at random between the old and the new
- Compare each checkout’s conversion rate
- Read the interval round the difference
Words you’ll meet
- Random split
- Sending each shopper to A or B by chance, like a coin flip, so nothing else decides who sees what.
- Control
- The version you have today, A, kept as the yardstick.
- Variant
- The changed version, B, that you want to judge.
- Conversion rate
- The share of shoppers who buy: 10% means 1 in 10.
- Interval
- The range the true difference most likely sits in, given the shoppers seen so far.
Step 1 of 5
The turnstile sends each shopper at random to A, the , or B, the : a .
Run day 1.