Set-Based Design

Set-based design keeps several solutions alive at once and narrows them as evidence arrives, instead of picking one early and iterating on it. It comes from Toyota’s product development, which Durward Sobek, Allen Ward, and Jeffrey Liker described as set-based concurrent engineering.

One example: A team with two credible approaches to a caching layer builds a thin version of each, measures both against the same load, and keeps the winner. A point-based team argues for a week, picks one, and finds out a month later.

We use it most to settle arguments. When a team is split, and neither side has evidence, “try both for a day and compare notes” costs less than the debate.

In the Do beat of a dojo challenge, that’s a standard tool. For user experience work, a design studio runs the same logic on paper: many sketches, then critique, then converge. The discipline is in the narrowing. Running several options takes time and money (though less than before AI), so decide up front what evidence will close each one.

This practice fits naturally with product experiments, which give you the evidence to narrow on.

Resources

• Durward Sobek, Allen Ward, and Jeffrey Liker, “Toyota’s Principles of Set-Based Concurrent Engineering” (MIT Sloan Management Review, 1999)
• Design Studio — set-based design for user experience
• Product Experiments — how to produce the evidence that narrows the set