Collaboration

A shared language between design and engineering

Tokens and component libraries help; habits at the design–engineering boundary are what keep work moving week after week.

Estimated reading time: 6 min read

Handoff is often described as design upstream and engineering downstream. In practice the current reverses whenever data shapes, latency, compliance, or performance push back against an interaction that looked simple in grayscale. Strong teams do not eliminate that friction—they build a clear way to discuss it early, before it becomes a surprise in QA.

Name things the way releases will recall them

A component named SearchPanel in Figma and FilterDrawer in the repo is a small mismatch that becomes a large tax when bugs are filed by people who heard the term in sprint review. Naming is not bureaucracy; it is shared memory.

  • Keep surface names aligned with PR titles when possible—you should not need a translation layer in chat.
  • Prefer stable nouns (“invoice row”) over release-specific slang (“phase-two tile”) that will go stale in six weeks.
  • Reserve “experiment” prefixes for genuinely disposable UI so engineers can guard them with feature flags without confusion.

Contracts at the boundary

Quality often slips between the design specification and acceptance criteria: microcopy omitted, spacing left as “system default,” hover states orphaned. Lightweight contracts—not legal documents, just bulleted intents—establish what must reproduce for something to merge. Designers, product managers, and developers can each check the same short list.

If it is not articulated at the boundary, assume it will be invented under pressure downstream.

Decision reversibility buys calm

Not every visual choice deserves a permanent component. Marking something as reversible—a layout experiment with a single consumer, a temporary campaign surface—gives engineering permission to implement quickly without baking it into primitives. The shared language here is explicit intent, not scope creep in disguise.

Ritual beats volume

Long comment threads rarely outperform a twenty-minute review with the file and the running build open. The goal is not consensus on taste; it is agreement on constraints: what must stay flexible, what is locked for accessibility, and where copy will change faster than typography.