Walkthrough testing

You must test the design before the thing you'd test exists, because by the time the building is real it's too late and expensive to fix. So testing is simulation — and the core technique is the walkthrough: deliberately moving a person through your design, step by step, doing what they'd really do, watching for what fails. You can't walk an abstraction through a plan, but you can walk Lakshmi (your persona is your test subject) — each persona becomes a probe you push through the design to see where it breaks.

Not glancing and thinking 'looks good' — a slow, cinematic re-enactment. Take one persona, one realistic scenario, and move them moment by moment, asking at each step: does the space support this or fight it? Walking Lakshmi through her morning surfaces three failures a static glance never would — the prayer dais sits too low for her knees, rising is hard, two cooks collide in the kitchen. None is visible in a floor plan; they appear only when a real person does a real thing across real time. The walkthrough also confirms successes (the path is clear, the morning light reaches the corner) — testing tells you what's working, so you know what to protect.

The imagined walkthrough you run yourself, in your head — instant, free, repeatable, but carrying your own blind spots and attachment. The real walkthrough brings the actual people to a physical prototype — the family points at the foam kitchen and says 'this feels small'; Lakshmi looks at the prayer corner and goes quiet. It catches what your imagination missed and corrects your bias, because they aren't attached to your design and know their lives better than your persona-of-them. The imagined is your fast private first pass; the real is the honest check.

A single walkthrough tests one moment, but a home is lived across a whole day, year, and life. Run many, choosing the scenarios most likely to break the design: the ordinary Tuesday (does daily life flow?), the festival lunch (peak load), the monsoon evening (everyone indoors, wet shoes), ten years on (aged knees, a child moved out, a grandchild). Choose the scenarios likely to expose flaws, not the ones that flatter the design — a test you design to pass tells you nothing.