6 CONCEPTS / ONE SHOPPING TASK

CRO, measurement and performance

Find friction through task tests and measure changes with clear definitions.

Choose patterns by need

A usability test finds problems; a randomized experiment can estimate impact. Neither provides a universal conversion promise.

Define audience, exposure, denominator and primary outcome before starting. Record what could invalidate the data.

Start with an observable shopper problem. Define a primary metric, a clear denominator and quality checks before launching an experiment.

4 test tasks

Measure whether pages become visible, respond to shoppers and keep elements stable. Evaluate speed on real devices and shopping tasks.

4 test tasks

Define each step as an observable event. Separate a checkout attempt from an accepted order and confirmed payment.

3 test tasks

Define eligibility before starting. Mobile, market and new shoppers can be useful segments, but post-hoc hunting creates false winners.

3 test tasks

Check observed participant counts against planned allocation. Unexplained imbalance can make an experiment unreliable.

3 test tasks

Measure more than completed purchases. A change increasing errors, returns or discount cost can harm the business.

3 test tasks

Data loss, incorrect variant allocation, returns/refunds, margin, accessibility and mobile performance.