The most expensive failure in software has no error message. It compiles, passes tests, ships on time — and solves a problem nobody had. Post-mortems across the industry agree with brutal consistency: the leading cause of product failure is not bad code but absent demand. Product design is the discipline that attacks exactly that risk, before the expensive part begins.

The arithmetic is what makes it non-optional. A research week and a prototype cost a rounding error against a quarter of engineering. If they change the plan even slightly — and they nearly always do — they have paid for themselves many times over.

Research that actually informs

Good research is not asking people what they want — users are famously unreliable prophets of their own behaviour. It is watching what they do: where the workarounds live, what the spreadsheet next to the official tool is compensating for, which step of the current process makes them sigh. A handful of well-run interviews and observations reliably surfaces the insight that reshapes a roadmap.

From research comes structure: the information architecture and user flows that decide, before a screen is drawn, whether the product’s shape matches the user’s mental model. Most "confusing UI" complaints are actually structure problems wearing interface clothes.

Prototypes: arguments you can click

A prototype settles in an afternoon what meetings cannot settle in a month. Watching five users attempt real tasks in a clickable mock exposes the truth about a design with zero engineering spent — the mislabeled button, the missed entry point, the step everyone skips. Each round of test-and-revise at the prototype stage deletes a future sprint of rework.

The output that engineering receives is then genuinely buildable: flows that survived contact with users, states and edge cases specified, a design system to compose from. Handoff stops being interpretation and becomes implementation.

Design as risk management

Strip the craft vocabulary away and product design is a risk instrument: it converts "we believe users want this" into evidence, at the cheapest possible point in the project. Companies that internalise this ship less software and win more — because everything they ship was tested as an idea before it was funded as a build.

Avoiding the wrong thing is the highest-leverage act in product development. That is the job.