Identify Existing Design Process

Mapping how UI decisions get made today — who designs, who reviews, what tools and handoffs exist — so the new system plugs into reality instead of fighting it.

Why it matters

A design system is adopted by people inside an existing workflow; if you don’t know that workflow, you’ll build a system that nobody’s process has room for. The existing-design-analysis tells you what the UI is; this tells you why it became that way — usually “no shared source, three designers, copy-paste handoff.” Naming the process gaps is how you predict where consistency will keep leaking after launch, and it directly shapes governance and contribution-guidelines.

How it works

Interview and observe, then map the current pipeline from idea to shipped pixel:

  • Roles — who actually decides color/spacing? A lead, every designer, or engineers at code time? Diffuse authority is the root cause of variant sprawl.
  • Tools & source of truth — Figma, Sketch, or “in the code.” Is there one file, or per-feature files that diverge?
  • Handoff — how does a design reach an engineer: redlines, inspect panel, or guesswork? Where is intent lost?
  • Review — is there a design/UI review gate, or does anything mergeable ship?
Maturity signalImmatureMature
Source of truthper-designer filesone shared library
New UI valueinvented per featurereuses a token
Review gatenone / ad hocdesign + a11y check
Handoffscreenshotslinked component specs

The output is a current-state process map plus a short list of gaps, each tied to a future governance mechanism (e.g. “no review gate → add a UI review step”).

Example

Discovery at one company: 3 designers, no shared Figma library, handoff via exported PNGs, no UI review on PRs. That fully explains the 14 button variants — each feature reinvented one. The fix list writes itself: one component library, an inspect-based handoff, and a lightweight PR check that new colors come from tokens. Adopting the system without these would let drift regenerate within months.

Pitfalls

  • Auditing pixels, not process — fixing the output while leaving the broken workflow means the variants grow back.
  • Assuming a process exists — small teams often have none; “informal” is the finding, not a failure to find one.
  • Skipping engineers — much UI is decided at code time, not in design tools; interview both sides of the handoff.
  • Mapping without acting — a process diagram that doesn’t feed concrete governance changes is decoration.

See also