The Rough Prototype, and the Car

You can now create high-fidelity prototypes really easily. People say that's great. Is it?

In product development, the early prototype, sometimes just paper, is rough. That did two jobs.

• It was much cheaper to build than a high-fidelity one.
• It only included the core elements.

AI has made high-fidelity prototypes very cheap too. The second job still matters hugely.

Because a rough prototype is so basic, it hides the distractions of a finished product, and the only things left to focus on are the core elements. Ask for feedback on a polished product and you hear "I don't like the font on this", about a feature that shouldn't exist at all.

Step back and ask what you are testing with your prototype. In many cases a high-fidelity one gets you worse feedback than a rough one.

Let's make it practical. Say I'm designing a new car and show you a super polished mock-up. You go straight to the look: the colour, the layout of the dashboard, the cupholder that's too small.

Very little of that matches what drives buying: the number of seats, the price, the size, the reliability, the engine. Some people will still give you feedback on those. But it's hard to get them there, because people react to what they see, and what they see is the details.

That's not an AI problem. When doing more is this easy, people think more is better, and that high fidelity is better.

When I work with a team as fractional CPO, it's one of the first questions I ask: what is this prototype testing?