As good developers* we test components, types, APIs, accessibility rules, and end to end journeys... and then we ship them. Unfortunately, most of the time the interfaces we ship never meet actual humans until they make it to production.

There’s a lot automated testing can’t tell us. Do users understand the language we’ve chosen? Does our information architecture match their mental models? Can they work out what to do next? Do they even notice the shiny new feature we’ve built?

Playwright is great, but completing a Playwright script isn't the same as completing a task - and just five lightweight usability testing sessions can uncover problems that thousands of automated assertions would miss.

Let's explore where automated testing ends and usability testing begins, how to run useful sessions without a research team or enormous budget, and how to turn what you observe into better interfaces—and better tests.