Skip to content
Engineering

A design system is not a component library

A Figma file full of components is not a system until it changes how a team decides. Here is the difference, and how we build something people actually use.

Noura Al-Harbi Head of Design
6 min read
Share

It is easy to confuse a handsome component library with a design system. The first is a set of ready parts: buttons, fields, cards, colours. A system is the agreement that explains when those parts are used, when an exception is refused, and how something new is added without collapsing consistency. A library without that agreement quickly becomes a museum the team visits in the first week and then forgets.

On projects where we have joined a team after a “system” launched and then went untouched, the file was full and new screens were still being invented from scratch. The cause was not visual laziness. It was the absence of rules that make compliance easier than improvisation. A successful system saves decision time, not only drawing time.

The library answers “what”; the system answers “why”

A component says: this is the primary button. A system says: use this button when the action is the main step in the context, and avoid it when two actions carry equal weight. Without the second sentence, every designer and engineer will choose according to the mood of the page, and disorder returns wearing tidy components.

So we start with principles before we stockpile states. One clear principle — clarity before decoration, or hierarchy before novelty — can replace dozens of repeated comments. Principles are not wall slogans. They are tools used in review: we ask whether a decision serves the principle or a momentary preference.

A system that cannot say no to a new plate is not a system. It is a filing drawer that only gets fuller.

Build what will be used this month, not what looks complete in a presentation

Early completeness kills small systems. A team running a site and a few internal pages does not need hundreds of states on day one. It needs the parts that actually repeat: type, colour, buttons, forms, cards, and notices. If that loop is done well, expanding the system becomes natural rather than a second, exhausting project.

A shared language between design and engineering

If component names differ between the file and the interface code, time is lost in translation. We unify naming early, document only the essential states, and attach each component to a live example in the product rather than an isolated picture. The live example reveals real limits: what happens with long copy, what happens in Arabic, what happens when a field fails.

That link makes the system part of daily work. An engineer does not have to guess intent, and a designer does not draw a fourth version of a button that already exists because the old one “felt far away” from the current screen.

What makes a team return to the system

Use is not imposed by an internal memo. It is earned when people find that the system shortens an argument they would otherwise repeat. In our practice, the following are what keep a system alive after the launch excitement fades.

  • Short principles written in working language, easy to call on in a review.
  • A limited set of components covering recurring cases, with empty, error, and loading states where they matter.
  • Clear guidance for exceptions: how a departure is requested, who approves it, and how it is recorded.
  • A simple contribution path, so nobody grows a secret branch in a private file.
  • Examples from real pages, in Arabic and English if the product is bilingual.
  • A short, recurring review, not a document written once and left alone.

Without a contribution path, an unofficial system grows beside the official one. Without a polite right of refusal, the official system swells until it is too heavy to approach. Balance matters more than abundance here.

Steps for turning a file into a living system

We do not ask teams to stop shipping until a constitution is complete. We ask them to close a small loop and then widen it. This sequence keeps momentum and stops the system project from swallowing the product project.

  1. List the components that repeated in the last two launches, and ignore, for this first stage, anything that appeared only once.
  2. Write only three principles, and test them against two recent decisions: would they have changed the outcome?
  3. Unify part names between design and code, and remove near-duplicate variants.
  4. Document use and non-use for each essential part in a short paragraph, not in a page nobody will read.
  5. Name a clear owner — one person or a pair — who can accept and refuse additions.
  6. Attach the system to a short review after each launch, and record what broke and what was missing.

After a few weeks the difference shows: discussion shortens, new screens start from a known base, and an exception becomes a conscious event rather than a quiet habit.

Screens remain, but they are no longer the only source of truth

A system does not abolish screen design, and it does not abolish visual judgement. It makes the screen a decision within a frame, not a full invention every time. When a new campaign asks for a different layout, the team can tell whether this is a legitimate extension of the system, a real need to open an exception, or simply a desire for novelty.

That distinction protects the brand over a longer horizon. Products that change quickly need a system more than they need a beautiful library on presentation day.

A system is measured by the decisions it spared you

The beauty of the file is not proof of success. Proof is that a busy team returned to the rules because they were faster than improvising. If your “system” is opened only in presentations, you have a library waiting to be turned into a living agreement.

We help teams build that agreement realistically, and attach it to the product rather than to slides. Talk to us if you want an honest review of what you have, and a move from components to a system people use.

Want an article that turns into work?

We will review your site or campaign and build a clear plan for the next step.