Design Principles
A short set of opinionated, prioritized statements that resolve design trade-offs — the tie-breakers a team reaches for when two good options conflict.
Why it matters
Principles earn their keep at the moment of disagreement. “Should this dialog block the user or warn inline?” is unanswerable from taste alone, but trivial under “Respect the user’s flow over system convenience.” Without them, decisions get re-litigated per feature and the loudest voice wins, which is how a product loses coherence. Good principles are prioritized, so when two collide you know which yields — unranked principles just move the argument up a level. They also feed governance: contribution reviews cite principles, not opinions.
How it works
A usable principle is specific, has a clear opposite, and implies an action:
- Testable, not a platitude — “Be simple” applies to everything and decides nothing. “Default to fewer options; reveal advanced settings progressively” is testable.
- Prioritized — order them so conflicts resolve top-down.
- Tied to consequences — each names what it sacrifices.
| Weak (platitude) | Strong (decides ties) |
|---|---|
| “Be consistent" | "Match platform conventions over brand expression" |
| "Delight users" | "Speed of task completion beats visual flourish" |
| "Be accessible" | "No information conveyed by color alone” |
Keep it to 3–5. More than ~5 and nobody recalls them under deadline, so they stop functioning as tie-breakers. They sit above the design-language foundations and shape concrete rules in guidelines and microcopy-guidelines.
Example
A data-heavy analytics product adopts: (1) Density before whitespace, (2) Show the data, chrome second, (3) Keyboard-first. When a designer proposes large card padding for “breathing room,” principle 1 outranks it — padding tightens. The principle did the deciding, not seniority. Contrast a consumer app whose principle 1 is the inverse; same question, opposite answer.
Pitfalls
- Generic values everyone agrees with — “intuitive, beautiful, fast” decide nothing because nobody argues the opposite.
- Too many — a list of 12 is a wishlist, not a ranked decision tool.
- Never invoked — principles that don’t appear in real design-review comments are decoration; cite them in contribution-guidelines reviews.
- No priority order — when “delight” and “speed” both apply, an unranked list just relocates the fight.