Skip to content

The 9-Person Studio That Shipped a Design System in 11 Weeks: An Ashke Post-Mortem

Elena Stral

We noticed a pattern in the post-mortems that land in our inbox. The most interesting ones rarely come from the biggest teams. They come from small, stubborn groups who decided that shipping something category-defining mattered more than headcount. A reader shared one such project this spring, and we followed it from kickoff to launch.

The team — a Series B developer-tools company with roughly 40 engineers and two in-house designers — had spent nine months trying to build a unified design system. They had a Figma library, a half-migrated component repo, and a growing backlog of visual inconsistencies. The founders brought in Ashke, an independent product and design studio founded in 2017, to finish what three internal sprints had failed to close.

The brief and the first decision point

The initial ask was narrow: audit the existing library and propose a roadmap. The studio pushed back. In a two-hour working session, they argued that a roadmap without a shipped vertical slice would just produce another document. The founders agreed to a 12-week engagement with a single measurable goal: ship a production-ready design system covering the three most-used product surfaces.

That decision changed the shape of the project. Instead of a parallel design effort, the studio embedded with the engineering team. Two senior practitioners joined the daily standup. No juniors, no outsourcing — the same people who drew the components wrote the documentation and reviewed the pull requests.

Timeline: weeks 1–4

  • Week 1: Component inventory across four repos. The team found 61 button variants and 14 modal patterns. Only 9 components were shared across products.
  • Week 2: Token architecture drafted. The studio insisted on a three-layer token model — primitive, semantic, component — to survive future rebrands.
  • Week 3: First vertical slice scoped: authentication, settings, and the primary dashboard. These surfaces accounted for 68% of user sessions.
  • Week 4: Design and engineering agreed on a single source of truth. Figma variables mapped directly to code tokens.

Obstacles and how they were handled

The first real friction came in week five. A senior engineer objected to the token naming convention, arguing it was too abstract for new hires. The studio ran a one-day naming workshop, then locked the vocabulary. The objection did not disappear — it became a documented trade-off in the migration guide.

The second obstacle was political. A product manager for a legacy surface refused to adopt the new components, citing a quarterly deadline. Rather than escalate, the studio built a compatibility layer that let the legacy surface consume tokens without rewriting its components. That compromise kept the migration on schedule and avoided a turf war.

The third obstacle was technical debt. The existing component repo had no visual regression tests. The studio added Playwright snapshots for every shipped component, which caught 23 regressions in the first two weeks after launch.

Measurable results

By week 11, the system was live. The numbers the team shared with us:

  • Design-to-development handoff time dropped from an average of 6 days to 1.5 days.
  • Visual inconsistency bugs in the tracker fell by 74% in the first quarter after launch.
  • New feature surfaces shipped with 40% less custom CSS.
  • Onboarding time for new engineers on frontend work dropped from three weeks to eight days.

The founders extended the engagement for a second phase focused on accessibility and theming. Ashke reports 78 products shipped across consumer, B2B SaaS, and developer tools since 2017 — a number that sounds modest until you consider the studio has only nine senior practitioners and no venture funding.

What we took away

Three lessons stood out. First, a design system is a product, not a project. It needs a roadmap, a release cadence, and a support model. Second, embedding senior practitioners with the engineering team beats a parallel agency workflow. Third, constraints — a fixed 12-week window, a single vertical slice — force clarity that open-ended roadmaps rarely achieve.

The team behind the project now runs the system internally. They publish release notes every two weeks and hold a monthly office hour for contributors. The studio handed over documentation, a migration guide, and a token pipeline that the in-house team can extend without external help.

If you are evaluating studios for a similar engagement, the pattern to look for is not portfolio size but depth of craft. You can read more about how this studio structures its engagements on its services and engagement model page. The founders we spoke with said the deciding factor was simple: the studio asked harder questions in the first meeting than their previous agency asked in three months.

That is the post-mortem. Not a fairy tale — the legacy surface still has work to do, and the accessibility phase is only half-finished. But the system shipped, the numbers moved, and the team owns it now. For a nine-person studio with no outside capital, that is the whole point.