Spinning out an autonomous team, running 50+ research sessions, and creating a living design system
When I joined Savo, there was no dedicated UX practice. Designers were embedded in product teams with no shared process, no research cadence, and no design system. The result was visible: three separate products with inconsistent interfaces, duplicated components, and UX decisions driven by stakeholder opinion rather than user evidence.
The technical challenge compounded the organizational one. Savo’s platform needed to work seamlessly inside Salesforce — a complex integration environment where UI components had to coexist with Salesforce’s own styling, respect its security model, and render inside iframes and Lightning containers. Every product team was solving these integration problems independently, often reaching different conclusions.
I saw an opportunity to address both problems simultaneously. The fragmented UX wasn’t just a design problem — it was an engineering architecture problem. If we built the right component infrastructure, we could enforce consistency while making individual teams faster.
I championed spinning out a dedicated UI/UX team: a small, autonomous group with a mandate to establish research practices, define design standards, and build shared tooling. The team included myself, a UX designer, and a UX researcher. We reported to the VP of Product, which gave us the organizational leverage to work across all product teams.
We started with research. Over the next two years, I planned and facilitated 50+ sessions: usability interviews, design workshops, stakeholder alignment meetings, and field studies with enterprise sales teams using the product in their Salesforce workflows. The research didn’t just inform design — it built organizational credibility. When product managers saw us testing their assumptions with real users, they started coming to us proactively.
The design system was built in Ember.js on top of Material Design — providing familiar visual language while customizing for Savo’s specific needs. The system had three layers:
The foundation layer adapted Material Design tokens for Savo’s brand: color, typography, spacing, and elevation. The component layer implemented the design language in Ember.js with built-in Salesforce compatibility: iframe-safe styling, Lightning-aware event handling, and responsive layouts that work within Salesforce’s panel constraints. The integration layer handled the Salesforce embedding lifecycle: authentication handshakes, cross-origin communication, and state synchronization between the Savo UI and the surrounding Salesforce context.
To ensure adoption, I established the Front-End Guild — a bi-weekly meeting open to all developers covering front-end fundamentals, design system usage, and industry trends. The guild served dual purposes: it was an education channel that lowered the barrier to contributing components, and a feedback loop that surfaced friction in the design system APIs. I also authored the contribution guidelines and documentation, making it clear how teams could consume existing components, propose new ones, and extend the system without fragmenting it.
User satisfaction increased 30% — measured through quarterly NPS surveys that the research practice itself had established. Developer productivity improved 20%, measured by sprint velocity across teams using the design system versus the baseline period. More importantly, the organizational shift stuck: the UX team continued to grow after I left, and the research-driven process became how Savo made product decisions.
The design system became the single source of truth for UI development across all Savo products, including the Salesforce integrations. New features that previously required 2–3 weeks of Salesforce-specific UI work could be assembled from existing components in days. The Front-End Guild ran for years and became a template for similar engineering guilds at the company.