Marcus Andrey Study
Ler em português Library

Design Systems · Step 1 · 6 min read

When you don't need one yet

A design system costs a team, time and patience, and it only pays back when work repeats for years. The signals that say not yet, start small or invest, what to do at each stage, and why the smallest useful system is usually a set of tokens and five components.

A design system is an investment, and like any investment it can come too early. Build one for a product that will change shape next month and you'll spend your time maintaining parts nobody uses. Wait too long and you'll spend it cleaning up drift. The skill is not "always build a system". It's reading the signals and building the smallest thing that helps right now.

What you'll be able to do after reading:

  • Explain what a design system costs, not only what it saves
  • Read the signals that say not yet, start small or invest
  • Choose what to do at each stage
  • Argue for starting small, with sources

1. What a system costs

The benefits are easy to list. The costs are what teams forget.

  • People. NN/g says creating and maintaining a system "is a time-intensive activity which requires a dedicated team", and suggests that "at a minimum, the team should include 1 interaction designer, 1 visual designer, and 1 developer" (NN/g, Design Systems 101, 2021).
  • Time to teach. The same article notes that "it takes time to teach others how to use the design system". A part nobody knows about is a part nobody uses.
  • Upkeep, forever. "Design systems require continuous maintenance and oversight." Every component is a promise to fix it, document it and keep it accessible.
  • Less freedom, early. A system makes the common thing easy and the unusual thing slower. That's the point once a product is settled, and a problem while it's still finding its shape.

2. When it's too early

NN/g names the clearest case: a system built "for a proof of concept or an initial prototype which is likely to change" is "probably not going to generate a desirable ROI in the near term". And it names the opposite: "design systems are best when the company foresees years of future, replicable design work."

So the two questions are: will this work repeat? and will it last? If the answer to either is no, a system is too early.

Brad Frost, who has built systems for many organisations, says the same thing from the other side. In his description of the design system ecosystem, most layers are optional, and he recommends starting simple and adding only what real needs ask for (Frost, The Design System Ecosystem, 2023).

Frost also gives the other side of the argument. Asked when to establish a design system, his short answer is "now", and he suggests riding on another project, like a redesign or a move to a new platform (Frost, Atomic Design, chapter 4). The two views fit together: start now, but start small, and grow only as the work repeats.


3. Reading the signals

SignalNot yetStart smallInvest
Products that share interfaceOneTwo, or a second one on the wayThree or more, or several platforms
People designing and buildingA few, in one room or one chatTwo or three teamsMany teams who rarely talk
How long the product will liveA prototype, a test, a campaignPast its first year, still changing fastYears of work ahead
Repeated partsLittle repeats yetThe same buttons and fields keep being redrawnDrift is visible and costs review time
Coming soonNothing that touches every screenDark mode, a rebrand, an accessibility auditA second brand, a new platform
What to doA shared file and a short style sheetTokens and the five most used components, one ownerA team, a roadmap, releases and docs
Read across each row and see which column describes you most often. Most teams land in the middle, and the middle is a good place to start. Tidewater, with two products and a dark mode on the way, is a clear "start small".Based on NN/g, Design Systems 101 (2021), and Brad Frost, The Design System Ecosystem (2023). Table: Marcus Andrey, CC BY 4.0.

No single row decides. A team of four designers with one product that will live for years might still be in "not yet". Two teams who have never met, building products that share a login screen, are already in "start small".

Tidewater has two products, three teams, years of work ahead, the same buttons being redrawn, and a dark mode coming for the console's night shifts. That's "start small", close to "invest".


4. What to do at each stage

Not yet. Keep one shared design file with your colours, type and the parts you reuse. Write a page of notes: "our primary blue is this, our spacing goes up in 4 px steps". That's not a system, and it doesn't need to be. It's the seed of one, and it costs an afternoon.

Start small. Pick an owner, even part-time. Turn the colours, type and spacing into tokens (Step 2). Build the five components used most often, usually button, text field, select, checkbox and a card or list row, in design and in code, with the same names. Write one page per component. Ship it to one product first.

Invest. A team, a roadmap, releases with version numbers, docs people read, a way to contribute. Steps 5 to 7 of this course, and the "For leads" lessons, cover this stage.

The order matters. A big team with no tokens and no adoption is a UI kit with a budget.


5. Open-source systems as a starting point

You don't have to build everything yourself. NN/g notes that "the more custom your design-system solution is, the more time and money it will take to implement", and that a custom system is worth it when the organisation "has particular needs that cannot be met by open-source design systems".

Public systems like Material Design 3, IBM Carbon or the GOV.UK Design System can be the base you theme with your own tokens. You get tested, accessible components; you give up some control over how they look and when they change. For a small team that's often a good trade.


6. Common mistakes

  • Starting a system because other companies have one.
  • Building thirty components before any product uses one.
  • Waiting for the "right time", while the drift grows.
  • Building a custom system when a themed open-source one would do.
  • Making the system a side project with no owner.

7. Vocabulary

Term Meaning
ROI Return on investment: what you get back for what you spend
Proof of concept A quick build to test an idea, likely to change
Owner The person accountable for the system, even part-time
Theming Changing a system's look through its tokens, without rebuilding components
Open-source design system A public system others can use and adapt, like Carbon or Material

8. Try it (no tools needed)

  1. Take the product you work on, or one you know well.
  2. Go through the figure row by row and mark one column per row.
  3. Count the marks. Then write the one thing you'd do this month, from the "What to do" row.

Reflection questions: Which signal will change first in the next year? What would move you one column to the right?


9. If you already practise

  • Write the case for or against a system as a one-page decision, with the signals, and keep it. When the signals change, you'll know why you decided.
  • If you have a system, check it against the stages: is it bigger than the product needs? A system can be too big as well as too small.
  • Before building custom components, try theming an open-source system for a sprint and list what it can't do.

Checklist

Learn more

Books, courses and videos to go deeper

Book links go to Amazon. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.

What a design system solves · Anatomy of a design system · Funding and staffing a design system team · A roadmap for the system · What a design system is

Saved only on this device.

↑ ↓ to move · Enter to open · Esc to close