Problem
8 Brands, 0 System
Hundred-year global coatings leader with no shared design language.
Enterprise · Design System · 2022
One logical framework underneath eight distinct brand expressions — operable by local teams across 83 countries who had never worked in a design system.
Problem
Hundred-year global coatings leader with no shared design language.
System
Semantic layer supporting brand variation without duplication.
Process
Built for teams who had never used a component library.
Outcome
Single framework, eight brand expressions, global consistency.
Artifacts from this engagement
Hempel Group is a 100-year-old global leader in protective coatings — ships, infrastructure, wind turbines, buildings. Eight distinct sub-brands. Eighty-three countries. One company with no shared design language between any of them. This engagement built the component library that changed that: a single logical framework underneath eight distinct brand expressions, operable by local teams who had never worked in a design system before.
Hempel's growth had been acquisitive — eight sub-brands built or bought over a century of operations, each with its own identity, its own digital infrastructure, and its own design approach. From a distance, this looked like diversity. From inside, it looked like eight separate companies that happened to share a parent organisation and a supply chain.
The cost was concrete. A campaign that needed to reach ship operators in Rotterdam, infrastructure project managers in Dubai, and wind farm developers in Denmark required eight separate design efforts even when the message was identical. Local marketing teams in each country were building from scratch every time. The design variance between what Hempel looked like in Germany versus what it looked like in South Korea was not a brand strategy — it was organisational entropy that had accumulated undisturbed for decades.
Global design systems fail when they're designed from abstraction — when an architect invents a logical structure without first understanding what actually exists. The Hempel audit catalogued every component in active use across eight brands before a single design decision was made.
Catalogued every UI component across eight sub-brand digital properties — 340+ unique components identified, mapped to functional categories, and analysed for structural similarity. The audit revealed that 83% of all components served identical functions across brands; only their visual expression differed. This finding reframed the entire engagement: the system needed to separate structure from skin at the component level.
Designed the three-tier token system: primitive tokens (raw values — colours, spacing, radii), semantic tokens (purpose-assigned — primary-action, surface-default, text-secondary), and brand tokens (the per-brand layer that overrides semantic tokens to produce each sub-brand's distinct appearance from a shared structural base). This architecture meant a single component could serve all eight brands simultaneously by swapping only the brand token layer.
Built 47 core components covering 94% of all identified use cases across the eight brands. Each component built at the structural level with full brand token coverage — meaning every component shipped simultaneously in eight brand variants. Accessibility compliance (WCAG AA) built into the component specification, not retrofitted. Documentation written for local marketing teams with no design background, not for design system engineers.
Phased rollout across eight brands with regional adoption support — working sessions with local teams in Copenhagen, Hamburg, Dubai, and Singapore to transition from existing systems. Established the design system governance model: a central system team owning the core library, brand leads owning brand token layers, and regional teams operating within the component framework with a structured contribution protocol for edge cases.
Every component built at the structural level — layout, spacing, interaction states, accessibility behaviour — with brand expression entirely managed by the token layer above it. Changing a brand's visual identity requires changing its token set, not rebuilding its components. The system supports brand evolution without component reconstruction.
Primitive → Semantic → Brand. Primitives hold raw values. Semantics assign purpose. Brand tokens override semantics per sub-brand. A single change at the primitive level propagates correctly across all 47 components in all eight brands simultaneously. Brand changes are localised. System-wide changes are atomic.
Documentation written for the actual users of the system — regional marketing coordinators in 83 countries, most without design backgrounds. Every component page includes a usage decision tree ("when to use this / when not to"), a brand variant switcher, and explicit do/don't examples. The system is usable without a designer in the loop.
Central ownership of the core component library. Brand leads own their token layers. Regional teams operate within the framework with a structured contribution protocol. No single team is a bottleneck. The system grows through contribution, not through central committee approval of every edge case. Governance designed to stay lightweight as the organisation scales.
We had tried to build a shared design language twice before and failed both times. The system always collapsed under the weight of eight brands trying to protect their own identity. Raghvendra solved that problem architecturally — by separating the structural layer from the brand layer, he made it technically impossible for one brand to compromise another. That architectural insight is why this system survived and the previous attempts did not.
The instinct in multi-brand system work is to invent — to design the ideal shared component from principles. The audit finding at Hempel inverted that: 83% of what existed already was functionally shared. The work was not creation but recognition — seeing the shared structure underneath the visual divergence and naming it. Auditing before architecting is not a phase to compress. It is the work.
The two previous attempts at a Hempel shared system failed because brand managers vetoed components that felt like compromises to their brand identity. The structure/skin separation resolved this politically by making it architecturally true: no brand ever shares visual expression with another. Brand identity is preserved at the token layer. When the architecture makes compromise unnecessary, the resistance disappears.
Design system documentation is almost always written by designers, for designers. At Hempel, the actual system users were regional marketing coordinators in 83 countries who had never opened Figma. Documentation written for them — with decision trees, plain language, and explicit examples — delivered three times the adoption of the previous system's technical documentation within the same timeframe.
Design systems die in the gap between "the core system is built" and "the org knows how to maintain it." At Hempel, governance was designed and communicated from week six — before the components were finished. Every stakeholder knew their role before the system launched. The federated model meant no single team became a bottleneck, which is why the system survived three acquisitions it was never designed for.
DesignOps 360 covers multi-brand system architecture — token systems, component libraries, governance models, and the documentation that makes them survive in organisations that don't have designers on every team. Transparent pricing from ₹8L.
Seven more case studies across enterprise logistics, fintech, urban mobility, practice building, and AI strategy — all built on the same structural rigour applied here.