Northwind's checkout had not been touched in three years. Conversion was 2.9 %, every team had a theory, and nobody owned the last step. I asked for the last step.

The question

What stops people after they have decided to buy? We watched forty sessions, read six months of support tickets and found three answers: the address form, the surprise of a delivery fee, and a payment page that looked like a different company.

What we shipped

• Address lookup instead of seven fields, in four countries
• The delivery fee shown on the product page, not at the end
• One payment page in Northwind's own design, with the three methods people actually used
• Northwind Pay: a saved-card wallet, launched one country at a time

The number

Conversion went from 2.9 % to 3.8 % in a year. Most of it came from the address form, which nobody had bet on.

The checkout is the whole funnel. Everything before it is a promise; the checkout is where the promise is kept.

What I would do differently

Ship the fee change first. It took us four months to agree on it and two weeks to build.