Challenges

The mobile app was losing trust: app ratings around 1.6–1.7★, slow 5–7s interactions, 15–20% failed transfers, and ~25% retention. Support was swamped; UI and processes differed across 16 countries, and there were no analytics to steer decisions. Organizationally, there was no design governance, unclear ownership, and slow release cycles.

The Process

Discovery & decisions. We instrumented analytics (Firebase/Mixpanel + Smartlook/Hotjar), ran interviews, and mapped jobs/flows to expose drop-offs (e.g., 65% bounce on key screens, 60% confirmation drop-off). A “rebuild vs. patch” workshop with stakeholders led to a full redesign decision tied to a unified design system.  

Design system & execution. We built a “bank-grade” component library (WCAG-ready, dark mode, motion guidelines), approved a new tone of voice via legal, localized for 8 languages, and A/B-tested microcopy (+25% funnel lift). State design and safe defaults were standardized for high-risk flows.  

Org & operating model. I hired the initial team in five weeks, introduced OKRs (e.g., ≥4.5★ rating, NPS ≥60), bi-weekly sprints, monthly reporting, and cross-functional reviews with Product, Engineering, Compliance, and Support. We launched in-app NPS/CSAT and real-time dashboards within 48 hours of go-live.  

Iteration & course-correct. An early, more playful direction under-performed in tests (“not bank-grade”); we pivoted to a more conservative, trust-first visual and content system—without losing clarity or speed.

‍

The Outcome

• App rating: 1.4–1.6★ → 4.8★ (iOS) / 4.6★ (Android) in 20 months.  
• NPS: 15 → 62; SUS: 35 → 85; rage-clicks: −68%.  
• Retention (30-day): 18% → 62%; DAU growth: +670%; DAU/MAU: 18% → 45%.  
• Transfer completion: 45% → 82%; bounce rate: 65% → 32%.  
• Support tickets: −65% ≈ €1.95M/yr saved; engineering maintenance: −25% ≈ €2M/yr saved.  
• Digital revenue share (HU): 15% → 45% (≈ €300M+ incremental digital volume).  
• Design-system productivity: ~€32M/yr equivalent gain from reuse/governance.  

‍

‍

‍