Skip to content
Back to case studies

Design Leadership / Org Design

Creating a centralized UX <> Engineering team

Restructuring a fragmented product organization through centralized design leadership: four value streams reorganised into one hub.

Role
Sr. Director of Product Design, ORFIUM
Timeline
2016 to 2024

The hardest design problems are not on screen. They are organizational.

CODA is an organizational-design initiative, not a product. As Senior Director of Product Design at ORFIUM, I restructured the product organization by creating a centralized design operations and alignment model that brought product designers, front-end engineers, and the design system team under one leadership. This is the story of why the old model failed, how CODA replaced it, and what it changed.

At a glance

4 to 1
Value streams to one hub
8
People (4 PD, 4 FE)
6
Ceremonies mapped to the cycle
6
KPIs, all improved

Co-led with the Head of Front-End.

The original structure

ORFIUM ran on a distributed, value-stream model. Four independent streams, each with its own embedded product, design, and engineering people.

Autonomy came at a cost. Each stream became a self-contained unit, and design practice siloed. Every stream grew its own patterns, conventions, and definition of quality. Duplicated components, inconsistent interactions, and a design system that existed in theory but was adopted unevenly. The model worked when the company was small, then the cracks showed as the product surface and the teams grew.

des pm eng Stream A des pm eng Stream B des pm eng Stream C des pm eng Stream D DS
Four disconnected value-stream silos, with an isolated design system in the middle.

The breaking point

A company-wide restructuring collapsed the four streams into fewer, broader teams, and the alignment between design, product, and engineering shattered.

Designers were reassigned, reporting lines changed, context fragmented. Shaped work entered build cycles with ambiguous scope and unresolved dependencies. Designers produced specs engineering could not execute, not from a skill gap, but because there was no shared meaning of "ready for development". Handoff quality depended on which designer you happened to work with. The design system team built components nobody adopted, because it had no seat at the product table.

The problem was not individual performance. It was structural.

Product X PD FE DS 3/3 roles filled Tool A PD FE DS 3/3 roles filled Product Y PD FE DS 3/3 roles filled Tool B PD FE DS 3/3 roles filled Product Z PD FE DS 3/3 roles filled Tool C PD FE DS 3/3 roles filled // fully staffed REORG Product X PD FE DS 2/3, DS missing Tool A PD FE DS 1/3, critical Product Y PD FE DS 2/3, FE missing Tool B PD FE DS 1/3, critical Product Z PD FE DS 2/3, DS missing Tool C PD FE DS 0/3, abandoned // roles missing
Fully staffed product teams before the reorg, teams with missing roles and reduced capacity after.

Designing a new model: CODA

Instead of every product owning its own designer and front-end engineer, I proposed one centralized, shared team that could support every stream dynamically. That team became CODA.

CODA allocated support based on actual priorities, company objectives, and the shaping pipeline, not a fixed headcount split. It was four product designers and four front-end engineers, co-led with the Head of Front-End. The design system team merged in too, forming a hybrid federated model: ownership stayed central, but contributions could come from anywhere.

VALUE STREAM A VALUE STREAM B VALUE STREAM C CODA PD PD PD PD FE FE FE FE DESIGN SYSTEM
CODA as a central hub distributing to the value streams, with a bidirectional relationship to the design system.

The design system inside CODA

Merging the design system into CODA solved several problems at once. The system team was no longer isolated or under-resourced. Designers and engineers could rotate between system and product work, which prevented stagnation. The system finally reflected real product needs instead of hypothetical improvements, and quality rose because contributions came from many contexts.

Operating inside Shape Up

ORFIUM used a customized version of Basecamp's Shape Up: six weeks of focused work followed by a one-week cooldown.

The embedded model was meant to align with this rhythm. Designers shaped work, engineers picked it up the next cycle, the system moved predictably. Once resources shrank, it broke. Work was not ready at the start of cycles, multiple teams blocked at once, and shaping outran the capacity to support it. CODA was the way to re-synchronise the organization with the Shape Up cadence.

6 weeks 2 weeks 6 weeks DISCOVER DELIVER SHAPE BET BUILD COOL DOWN SHAPE BUILD
The Discover and Deliver lanes running in parallel.

New ceremonies

To keep CODA aligned internally and with the wider org, six ceremonies were introduced, each mapped to a point in the Shape Up cycle.

CeremonyCadencePurpose
CODA StandupsDaily, 15 minSurface blockers, share progress, coordinate across active projects
Design and FE PairingWeeks 2 and 4Structured designer and engineer collaboration so design intent translates accurately
Mid-Cycle ShapingWeek 3Ensure upcoming work is well defined, shaping runs in parallel with the build
Readiness ReviewWeek 6Evaluate every deliverable against the definition of "ready for development"
Cooldown PlanningCooldown weekPrioritise system work, shaping, process improvements, and tech debt
Design System ReviewRecurringReview contributions, component health, and adoption

Measurable impact

CODA's success was tracked through six indicators spanning operational health and team effectiveness. All six moved in the right direction.

Cycle readiness improved

Shaped work entered cycles with clearer scope and fewer unknowns, reducing mid-cycle disruption.

Resource utilization stabilized

Centralized allocation removed idle capacity and the chronic over-commitment of the old model.

Rollovers decreased

Fewer projects carried into the next cycle, a sign of better scoping and more realistic commitments.

Cross-team alignment improved

Shared ceremonies and one leadership structure cut miscommunication and conflicting priorities.

Design system activity increased

Contributions grew once the team was in the product cycle, and component adoption rose.

Employee engagement rose

Satisfaction improved as people gained clarity on their roles, impact, and growth.

Reflection

The user was the team itself.

Restructuring a fragmented org took the same skills I use in product design: deep research, ruthless prioritisation, clear communication, and iterative validation. The most important lesson was that organizational design is never done. CODA was not a one-time fix, it was a living system that needed continuous shaping, just like the products it supported.

CODA is part of my work at ORFIUM, alongside the Ictinus design system and the ORFIUM platform.