Green does not mean correct. Red does not always mean broken. You are the test engineer: manipulate timing, isolation, retries and assertions, then run an entire test suite. Learn to tell an actual product defect from a test that lies.
Choose what is wrong with this imaginary checkout feature. Then design how the test will observe it.
Configure your test on the left and run the experiment to collect observations.
A test checks that a click occurred, but never verifies that an order was saved. It reports success while the user journey is broken.
A correct asynchronous result arrives after a brittle fixed delay or an immediate check. Waiting for the actual outcome would avoid the race.
A retry can recover from a transient problem. Playwright classifies tests that fail initially and pass on retry as flaky—not simply passed.
Two tests mutate the same account or inventory record. Parallel runs become order-dependent, even when every individual test is well written.
A customer pays successfully in production. In CI, the test sometimes times out waiting for a confirmation message. Enabling one retry makes the pipeline green. What is the best next step?