🔒 Protected case study
The Raisin brand evolution case study is password-protected. Enter the password to continue.
Solo designer embedded in the B2C Growth and Engagement tribes — reporting to VP Design, taking project brief from Head of Product.
The starting point
Identified Gap 01
Diagnosis: Web showed a catalogue, not holdings. Mobile and email were different products — no single home for wealth.
Diagnosis: On web, customers met a catalogue — not their holdings across categories. On mobile, the same job met a different product and the emails didn't match visual language. One customer, no single home for their wealth.
Identified Gap 02
Diagnosis: Koto handed off a static identity — not a live, multi-market product brand.
Diagnosis: Brand agency Koto delivered a static identity. After handoff, Raisin design had to scale it across product experience — guidelines existed; a live, multi-market product brand did not.
Identified Gap 03
Diagnosis: Ad hoc craft and weak research rhythm — every initiative paid a tax.
Diagnosis: The rebuild needed a team that could move as one system. What existed was ad hoc — loosely shared craft conventions, a weak research rhythm, and siloed workflows — so every initiative paid a tax, including later work on AI.
Closing the gaps
Two programmes, run in parallel — not one tidy sequence.
Practice = outcomes · Enablement = capability. Brand evolution and team rituals — in parallel.
My remit: close the three gaps — brand evolution into live product (absorb the agency handoff and carry it into the surfaces customers use; redesign web, mobile, and communication into one home for their wealth), and enablement in parallel (ways of working, research and critique rituals, craft standards, and the AI programme) so the team could sustain it. Practice = customer outcomes. Enablement = team capability. I where I owned it; we at tribe scale.
Practice
Research-led decisions, shipped screens, measured outcomes.
Enablement
Rituals, research OS, docs, tooling, coaching — I built; the org adopted.
Impact: NPS, task success, adoption, AuM, and retention all moved on the customer side — alongside research rituals, playbooks, AI workflows, and coaching the org adopted. Measured honestly in Proof; not everything lifted.
Catalogue landing, not a home for wealth. Research → trade-offs → Wealth Hub MVP across 12 markets.
The web dashboard still behaved like a product catalogue — deposits first, not the customer’s holdings. This chapter turns that surface into a single home for wealth across savings, investments, and retirement, and carries the new brand into a live, multi-market product.
I led Wealth Hub research, IA, and the design bets. We shipped MVP across 12 markets.
Baseline research · 🇩🇪 · One week · 4× designers
We sat with B2C customers and asked what Weltsparen actually felt like after login — what they needed, what was missing. Not a polish brief; a test of whether a redesign was justified.
Sample set · 8 interviews (±2) · Raisin’s existing B2C archetypes
Customer problem
A catalogue landing, not a home for their money. Savers couldn’t orient to what they held across savings, investments, and retirement.
Orient — consolidated view across asset classes · Monitor — total wealth at a glance · Act — maturities and pending tasks surfaced
Co-creation · synthesis · two-week sprint
3× designers collaboration · 1 week
We spent one week clustering baseline findings into themes — then translating each theme into How-Might-We statements the whole tribe could ideate against.
Most desirable · Most feasible
How might we make what’s held at each bank — and how much — visible at a glance?
How might we bring ease in navigating across active and available products?
How might we balance clarity in one view for what’s invested vs transactional?
Competitive patterns — what peers showed on landing vs drill-down.
Jobs to be done and design principles from the brief.
Critique with PM, HoD, and engineering — one hierarchy taken to research.
Variation A explored in critique · B shipped — overview → asset class → bank
Wealth Hub · ideate
I facilitated ideation from competitive reads through wireframe critique — aligning the tribe on one holdings-first hierarchy before concept validation.
Design conflict · Product vs Design
Position: A flat list reads as a catalogue — the problem customers named. Hierarchy orients by what they hold.
Settled: Design critique — hierarchy won over flat catalogue list.
Product design strategy · trade-off 01 · validated with customers
Several landing decisions still needed customer evidence — not critique. I led Raisin’s first international study on allocation visualization: treemap vs donut.
Sample set · 12 participants · 2 prototypes · PRE + POST-KYC · 🇩🇪 🇳🇱 🇪🇸 🇬🇧 🇫🇷
Key trade-off
Position: Overview-first landing — detail on drill-down.
Research conclusion: Donut beat treemap — allocation legible at a glance, route into account actions.
Outcome: Donut + level in/out shipped in MVP
Variation A explored in study · B shipped
Also in the study
Asset-class hues · savings-only remaps · donut child shades — 12 markets, light + dark.
Listing row anatomy — hierarchy, metadata, and actions aligned to the component library.
Final interface — colour map and components applied across the landing experience.
Wealth Hub · systems
The hierarchy needed a colour system, not just a layout. I gave each asset class a distinct, accessible identity across 12 markets and both themes — savings blue, investments amber, retirement green — with rules for savings-only markets, documented so design, engineering, and markets shipped from one source.
12 markets · 3 market-type rules · light + dark token specs
Type, colour, layout, and components — one system for twelve markets. The colour map settled brand debate; typography, spacing, and components shipped in parallel with Foundation-fix.
After MVP launch · against targets declared before design
Customers used Wealth Hub as a home for their wealth, not a product catalogue.
Brand Evolution survey · ~1,900 respondents · DE / NL / ES / PL
Adoption target · 3 months post-launch · Met
Task completion target · Not systematically measured pre-ship
Agree they can better understand and manage their wealth
Dashboard / design NPS (DE +35 · NL +23 · ES/PL +53)
Business context: AuM reached ~€77bn by end 2024; the dashboard migration protected NPS and HNW loyalty through the rebrand.
What customers praised wasn’t polish — it was the structure:
“Everything works great. I especially like the drill-down graphics — Overview → Fixed deposit → individual bank. That’s how every website should work.”
“Finally a dashboard that shows my full portfolio in one place — much clearer than before.”
MVP live — measure every release. In parallel: Ways of Working, prototyping standards, and a listening loop.
With the MVP dashboard live, we entered monitoring mode — launch surveys, direct customer feedback, and a clearer read on what each release moved. In parallel, I enabled the design team: Ways of Working, Principles of Prototyping, and a shared measurement frame so product learning and craft could compound together.
I owned post-MVP discovery and feedback-led prioritisation. I led design enablement and measurement. We shipped with engineering.
Post-MVP · listen & ship
The MVP created a listening loop — not a second redesign brief. I turned launch signals into a prioritised release queue and measured each shipment on its own terms.
Shipped in sequence:
Release 01 Filtering
Category, deposit, and country lenses with availability.
Impact: All-deposits view led adoption.
Release 02 Zero-state
Zero-state A/B — donut carousel vs empty, two markets.
Impact: No conversion lift — visual hypothesis without enough upstream qual. Tightened how we form conversion A/Bs after. Shipped for clearer empty-state UX.
Release 03 Raisin Wrapped
Year-in-review for savers — design-initiated, no PM brief.
Impact: Strong reach/completion · soft AuM (not significant). Next: bridge reflection → action (renew / add funds).
Release 04 Email system
Atomic email system — transactional and marketing, Bloomreach-ready.
Impact: Master-template concept · spans every marketing & transactional use case.
Enablement
Every project restarted file structure; every prototype landed at a different fidelity. Figma Ways of Working standardised the files. Principles of Prototyping standardised the conversation — what to build, at what fidelity, and why.
WoW: adopted team standard · still how every project starts
PoP: shared language for prototype fidelity · still how we critique
Research · operating model
User studies had been rare and German-only. Wealth Hub changed that — multi-market interviews, synthesis, stakeholder alignment. The MVP launch survey showed structured feedback at scale. I authored the research operating model: when to use interviews vs surveys, how participants are recruited through Bloomreach, hypothesis-first briefs, synthesis in Miro, and the compliance layer so legal and markets could sign off once, not per study.
6-step workflow · Bloomreach recruitment · Confluence playbook
Mobile stalled after the tribe split. Desk research → prototype → Maze → ship — plus Cura, the horizon.
The tribe split in two, the mobile redesign had stalled for months, and nobody had written down where the product was going. I took mobile end-to-end — desk research, in-flow prototyping, Maze validation, Cura vision — and raised the team's prototyping bar.
I led mobile IA + Maze validation. I built the in-flow prototype with One Data. Shipping now.
Desk research
Legacy mobile: 42% / 25% / 33% interaction split — no pattern worth copying from web.
Web Wealth Hub earned its structure through customer research. Mobile didn’t have that yet — only a stalled redesign and pressure to copy across. Desk research on legacy interaction patterns (42% / 25% / 33%) showed why that shortcut failed: real usage didn’t match web’s model, and no single mobile pattern had won.
Legacy mobile · interaction data · pre-ideation baseline
Web ↔ mobile parity
Inherit the Wealth Hub system. Redesign the mobile surface — don’t redo twelve markets of research.
Web had earned its structure through research and shipped across 12 markets. Mobile’s job wasn’t to repeat that work — it was to translate it. After desk research ruled out replication, I aligned the tribe on a simple rule: inherit the Wealth Hub concept and system, redesign the mobile surface. That unblocked ideation without discarding what was already validated.
Inherited: hierarchy + colour system · Redesigned: tabs + Home
Mobile · ideate
Grounded in the mobile dashboard study brief: asset-on-platform clarity, Portfolio for owned products, and legible primary sections. Neither concept was pre-selected — both went into validation as Hypothesis A and B.
Concept validation · Hypothesis A vs B
Two clickable prototypes — same wealth jobs, different home structure
New mobile IA · asset on platform · insight on Home
Four tabs — Overview, Portfolio, Products, Wallet. One Data wired into a live prototype.
Overview, Portfolio, Products, and Wallet — parity with the web Wealth Hub. Home surfaces total assets and performance; Portfolio holds active products. I wired One Data personas into a live prototype so reviews ran on realistic mobile, not flat mocks.
Key trade-off · insight
Position: Thin analytics as a primary tab erodes trust on mobile — show wealth and performance where people land; put active holdings on Portfolio.
Maze validation · Unmoderated study · Two-week
Tabbed IA won — 92.9% vs 71.4% task success. Open risk: pending products (42.9%).
Two-week unmoderated study on the live prototype — portfolio-card vs tabbed IA. Research goal: confirm asset-on-platform Home improves wealth clarity, the Portfolio tab still works for owned products, and primary sections stay legible. Portfolio-card won — 92.9% task success vs 71.4%. Open risk: locating pending products 42.9%.
Shipped · mobile dashboard
Maze-validated IA — rolling out light and dark across core B2C markets.
The IA that survived Maze is the mobile dashboard: Overview, Portfolio, Products, and Wallet — one system on the Wealth Hub colour map, light and dark, aligned with web. Launching across core B2C markets on the same spine we validated there.
Early signal is strong — app-store reviews are already positive. Design NPS and structured task metrics are still being recorded; this is fresh post-launch, not a closed impact story.
Maze-validated · shipping now · early reviews 5/5
What reviewers praise first isn’t polish — it’s structure and overview:
“Clear and informative — exactly what I need at a glance.”
“Clear and easy to overview — everything in one place.”
“Simple setup, easy deposits, and a good overview of my money and the returns I’ve earned.”
“Easy to invest and move money between products and my linked account — best way to stay in control of savings.”
Shipped IA — drag to compare light and dark · switch tabs
Design vision · VP mandate · I authored the narrative
VP initiated the vision exercise. I owned the work inside it — Gather, Envision, Plan — and authored the 8-slide Cura deck: from reporting the past to helping savers decide next.
Gather → Envision → Plan · 8-slide narrative authored solo
Workshop to 180-day plan. Four production tools, eight designers coached — Showcase + Clinic until it stuck.
What started as a prototyping workshop became design's 180-day operating plan. As one of four org-wide AI Champions, I built four tools for daily production use, championed the operating plan, and led Showcase and Clinic until the workflows stuck.
I built the toolkit. We ran the programme; I championed design's slice.
Origin
One of four Champions — Lab, Showcase, Clinic, 180 days.
VP launched org-wide AI enablement. As one of four AI Champions, I wrote design's operating plan and ran adoption — Lab kick-start, weekly Showcase demos, and Clinic office hours I led. Coached eight designers on Cursor foundations first, then autonomy on live work; the programme was mostly coaching, the toolkit supported it.
1 of 4 AI Champions · Kick-start Lab → Showcase + Clinic · 180 days
Design × Cursor toolkit
Established an infrastructure every designer could run.
Setup
Install Cursor, connect Figma MCP, and Git basics to get a repo running.
Rules
Shared .cursorrules — tokens, typography, accessibility, and One Data on every prompt.
Skills
Repeatable pipelines — design QA, annotate, and variant workflows.
Subagents
Specialist modes — component architect, design systems, state and theme sync.
Commands
One-shot shortcuts — clean, polish, build, sync, responsive, a11y.
Workflow
Iterate until clean — draft → polish → micro-fix → responsive → motion.
Project Registry · Examples · Confluence
Lab · 4-week sprint
Design QA, copy, typography, annotation — four tools in daily production.
A structured build sprint — Design QA, copywriting, typography, and prototype annotation — all in production use.
Coaching · applied
Eight designers — Cursor foundations to shipping on live work.
Coached eight designers from Cursor foundations to autonomy on live work — setup, prompts, reviews, and shipping. One example: a cumulative earnings exploration with Deposits, vibe-coded in the open. The feedback-loop plugin made reviews native — Figma-style comments on the prototype (press c), synced as GitLab issues with file and line metadata, so PM notes landed on the exact line of code in Cursor.
8 designers · foundations → live work
Cadence
Open to join — drop in, unblock, and go.
Designers pushing MRs for UI fixes through Storybook.
Registry admin — kept the live catalogue current as we shipped.
Evangelized AI workflows and kept documentation alive for the team.
Proof
Some moved. Some didn't. I measured both.
Some numbers moved (NPS, task success, loyalty). Some didn't: zero-state — no conversion lift; Wrapped — soft AuM, not significant. I measured both.
Practice
Customer outcomes.
Enablement
Team capability.
The brand, built out
From static guidelines to a living system
— across every touchpoint customers actually use.
