Design Tokens: How We Keep Brand Consistency Across Every Screen
Technical

Design Tokens: How We Keep Brand Consistency Across Every Screen

Technical May 2026 · 7 min read

Design Tokens: How We Keep Brand Consistency Across Every Screen

One color value defined in a hundred places is a maintenance disaster waiting to happen. Design tokens solve this — and they're the reason our clients' brands stay coherent as their products scale.


A client comes to us six months after launch. They have a web app, a marketing site, a mobile app, and a component library built across two different teams. The primary brand red is #E5342A on the app, #EF4444 on the marketing site, and #E63946 in the component library. Nobody noticed until an investor pointed it out during a pitch deck review.

This is not a rare story. It's the default outcome when design and engineering teams scale without a shared source of truth. Design tokens are that source of truth.

"A design token is not a variable. It is a decision — named, versioned, and shared between every surface your brand touches."

What a Design Token Actually Is

A design token is a named value that represents a design decision. Not #ef4444 — but --ds-brand-primary. The distinction matters enormously.

When you hardcode a hex value, you've made a decision with no record, no name, and no path for updating it across your entire system at once. When you define a token, you've created a contract between design intent and implementation. Change the token, and every surface that references it updates simultaneously.

Tokens can represent anything: colors, type sizes, spacing, border radii, shadow depths, animation durations. Any value that should be consistent and intentional across your product is a candidate for tokenization.

Token Structure — design-tokens.css
/* Raw tokens — atomic values */
--color-brand-500: #ef4444;
--space-base:       8px;
--radius-md:        10px;

/* Semantic tokens — intent-named */
--color-action-primary: var(--color-brand-500);
--space-section-gap:    calc(var(--space-base) * 10);
--radius-card:          var(--radius-md);

Two Layers — and Why Both Matter

A well-built token system has two layers. Most teams build only one, and it causes problems at scale.

Layer 1: Raw tokens are atomic values with no semantic meaning. They're your palette — every color your brand system is allowed to use, every spacing increment, every radius value. They answer "what values exist?"

Layer 2: Semantic tokens reference raw tokens and describe intent. --color-action-primary references --color-brand-500. Semantic tokens answer "what is this value for?" This two-layer structure is what makes theming — including dark mode — genuinely maintainable rather than a hacked mess of overrides.

Layer Token Value Used By
Raw --color-brand-500 #ef4444 Semantic tokens only
Raw --color-neutral-50 #f4f4f4 Semantic tokens only
Semantic --color-action-primary var(--color-brand-500) Buttons, CTAs, links
Semantic --color-background var(--color-neutral-50) Page bg, muted surfaces
Semantic --radius-card var(--radius-md) → 10px All card components

Tokens Across the Full Stack

Here's where teams get it wrong: they define tokens in CSS and stop there. But the same product might render on web, iOS, Android, email, and internal tooling. Each platform has different style primitives. Design tokens only solve the consistency problem if they travel to every surface.

Tools like Style Dictionary allow you to define tokens once in a platform-agnostic format (JSON or YAML) and compile them into CSS custom properties for web, Swift constants for iOS, Kotlin values for Android, and JSON for design tools like Figma simultaneously. One source of truth, every platform.

tokens.json — Platform-Agnostic Source
{
  "color": {
    "brand": {
      "500": { "value": "#ef4444" }
    },
    "action": {
      "primary": { "value": "{color.brand.500}" }
    }
  },
  "spacing": {
    "section": { "value": "80px" }
  }
}

What This Means in Practice

When a client's brand evolves — and it always does — a team without tokens faces a project: hunt through every file across every platform to find and update hardcoded values. Miss one, and the brand is inconsistent. Miss several, and the product looks like it was built by different companies.

A team with a token system makes a single change to one value in the token file. Everything updates. The marketing site, the app, the component library, the email templates — all of them. A brand refresh that would take weeks takes an afternoon.

This is not a technical nicety. It's a business capability. The ability to move your brand quickly without breaking things is a genuine competitive advantage for any company that intends to grow.

01

Single source of truth

Every design decision lives in one place. No conflicting values across platforms.

02

Effortless theming

Light, dark, and brand variant themes become a matter of swapping semantic token values.

03

Designer–developer parity

Figma tokens map directly to code tokens. What designers spec is what gets built.

How We Implement Tokens at Nanite Studios

Every product we build starts with a token audit or token architecture depending on whether we're working on an existing system or starting fresh. We define raw tokens first, semantic tokens second, and only then do we begin building components that consume them.

For web projects, we use CSS custom properties with a Tailwind v4 bridge so tokens are available both as raw CSS variables and as utility classes. This gives engineers flexibility without ever touching a raw hex value in a component file.

For multi-platform products, we integrate Style Dictionary in the build pipeline. Tokens are version-controlled alongside code, reviewed in pull requests, and documented in a living style guide that both design and engineering maintain.

The result is a product that can scale from a single marketing page to a multi-platform application without ever losing brand coherence. That's not a happy accident. It's the system doing its job.

From the studio

Explore our work

See how we apply these ideas in live product design, brand systems, and engineering case studies.