Transforming a corporate welfare portal from a confusing catalog into an intelligent, personalized benefit assistant.
Corporate welfare portals manage enormous catalogs (gym memberships, travel packages, language courses, pension funds) yet most users default to supermarket vouchers. The agency asked me a consuting to redesign the service discovery experience, shifting the portal from a phonebook of services to a personal consultant that recommends benefits based on the user profile and credit balance.
Up to 40% of assigned welfare credit goes unspent, not from lack of interest, but from sheer disorientation.
The problem: infinite benefits, zero guidance
Three structural issues undermined the experience: choice overload (infinite benefits with no filtering or guidance), confusing terminology (voucher, giftcard, buono, rimborso used interchangeably for completely different mechanics), and low engagement as a business problem — unspent credit is lost value for both the company and its employees.
Research: data over assumptions
With one week, I ran a compressed research sprint: a quantitative survey (N=27 real users) and 2 rounds of usability tests (N=10, mid-age 36-45 that use welfare portal). The survey covered profiling invasiveness, terminology clarity, navigation priorities, notification preferences, and cross-selling reactions. I segmented by life stage (not just age) which shaped the design more than any other variable.
What the data confirmed
These became the non-negotiable pillars of the design:
- 85.2% want their credit balance always visible, on every screen.
- 73% prefer transparent copy explaining why data is requested, over generic personalization messaging.
- 63% accept 3-4 profiling questions at launch. No fewer, no more.
- 33% identify speed as top priority — benefit activation in max 3 taps.
- 52% tolerate at most 1 notification per week.
What the data challenged
- 26% reject profiling entirely. Not a bug, a constraint. The portal must work at 100% without personalization.
- Senior users (46+) tolerate at most 1-2 questions, versus the 3-4 majority. A single flow would alienate them.
- Cross-selling irritates users, not the content, the context. Pension fund suggestions while activating an Amazon voucher feel like misplaced ads. Decision: below-fold only, never inline.
Three journeys, not one
A single onboarding flow would alienate a significant portion of users. I designed three adaptive paths:
1. Journey A – The Deep Profiler (~40%). 90-120 seconds of life-stage questions. Fully personalized homepage.
2. Journey B – The Fast User (~35%). 1-2 quick questions. Partially personalized, under a minute.
3. Journey C – The Explorer (~25%). Skips profiling. Standard homepage driven by peer popularity. Still functional, still useful.
Comparative journey
Profiling without being invasive
The profiling does not ask who you are.
It asks what you need.
Every screen is skippable, and every question includes a Why are we asking link, because 73% of users respond better when the reasoning is explicit.
- Screen 1: Interests (multi-select, optional). 6 macro categories. Invasiveness: 1.85/5. Safest question, comes first.
- Screen 1.1: Related categories (multi-select, conditional). Social proof-driven suggestions based on Screen 1 selections.
- Screen 2: Who benefits (multi-select, optional). Just me / Partner / Pets / Children / Elderly parents. Invasiveness: 2.81/5. Highest in the flow, so it comes after the user has already invested.
- Screen 2.1: Conditional detail. Appears only based on Screen 2 selections (e.g. children age unlocks daycare, school, summer camps).
- Screen 3: Timing (optional). Now vs. over time. Orders homepage by perceived urgency.
The age-agnostic principle
The system never asks for age. A 28-year-old and a 38-year-old with two young children have the same needs on the portal. Age adds no information not already captured by life stage.
Show value before asking for effort
Before profiling, a screen shows concrete examples of what the user can do with their credit (groceries, cinema, fuel, dining). The profiling flow had a 57% completion rate in initial tests.
Users did not understand what they would get in return: this screen solves the problem upstream by showing value before asking for effort.
From profiling data to recommendations
The algorithm is simple, transparent, and explicitly balances user goals with business objectives:
40% Explicit profile match. Direct answers from profiling. The most reliable signal because it is intentional.30% Behavioral patterns. Clicks, saves, activations over time. Users who skip profiling still improve their feed.20% Peer group popularity. Activations from similar profiles, not the global average. This is why Popular is the first homepage section for non-profiled users.10% Business margin. A slight boost for company objectives (e.g. pension fund promotion). Low by design, never overrides relevance. The 10% is not a dark pattern -- it is transparency.The solution: badges, onboarding, and clusters
The 4-Badge System
Four visual badges (Buono, Rimborso, Sconto, Gratis) replace confusing terminology.
In a glance: does this consume my credit, require upfront payment, offer a discount, or cost nothing?
Onboarding: three screens, three problems
Three skippable splash pages, each solving a specific research finding.
- Screen 1 (clarity): 57% do not understand badges on first exposure: this screen maps them visually.
- Screen 2 (speed): 33% worry about slow processes 3 clicks, no forms.
- Screen 3 (privacy): 26% resist profiling three explicit guarantees about anonymity, control, and optionality.
The order matters: clarity and speed first, then ask for trust.
Homepage anatomy: a proposal, not a catalog
Every section position maps to a research finding. The page tells a story from universal to personal:
Credit balance: always visible (85.2% request).
Benefit type filters: horizontal navbar filtering by mechanism (Vouchers, Reimbursements, Discounts, Free), not topic.
Scelti per te (For you): personalized section, only after profiling; non-profiled users see Popolari first without perceiving anything missing.
Popolari (Popular): social proof as navigation for undecided users.
In scadenza (Expiring soon): Real urgency, not manufactured (up to 6 cards sorted by deadline).
Novità vicino a te (New near you): geolocated benefits, only with location permission.
Categories: macro-area browsing, works for all three journeys. Inizia bene l anno — seasonal section for long-term benefits (pension, training, preventive health).
Iterations from testing
Usability testing
Badge comprehension test, profiling flow test, benefit activation test. Results produced 4 concrete design iterations. No decision was made on aesthetic preference or convention alone.
Each iteration traces to a specific test finding:
- Badge size and positioning. Colors were memorable, badges were not. Scaled from 100x24px to 112x26px with box-shadow to separate from photographic backgrounds.
- Redundant micro-copy removed. 71% judged labels like Non usa credito redundant against the badge. Both lines stripped. Every card element must earn its space.
- Font size increased. Explicit user feedback: the text below the title is too small. Uniform +2px across the hierarchy (title 14 to 16px, supporting 12 to 14px, meta 11 to 12px). Accessibility fix for the 46+ demographic.
- Multi-select made explicit. 3 out of 5 users treated multi-select as radio buttons. Added subtitle Select one or more options and visible checkboxes. One line of text, zero architecture changes.
Design system
Typography: Onest sans-serif. Bold for titles/badges, Medium for pricing, Regular for descriptions. Min body 14px on mobile.
Color: Tailwind palette with functional role: Indigo (credit), Teal (reimbursement), Amber (discount), Fuchsia (free). Contextual: Orange (expiring), Rose (nearby), Blue (popular). Each color always maps to one meaning.
Spacing: 4px base grid, every value a multiple of 4. Spatial hierarchy mirrors informational hierarchy.
Iconography: Google Icons, outline style.
Photography: editorial macro, single subject, neutral background. No logos or brand elements — the image communicates emotional context, the badge handles information.
Next steps & KPIs
To measure the success of the redesign post-launch, I defined strict KPIs:
- Badge Comprehension: Target >85% (measured via in-app survey after the first session).
- Time-to-first-benefit: <90 seconds for the core 36-45 demographic from login to activation.
- Credit Utilization: Increase spent welfare credit to 75% annually.
Conclusion
Assigning €1,200 of welfare credit is useless if the user doesn’t know how to spend it.
The true business value generated by this redesign is transforming a forgotten budget into a perceived daily value, proving that solid, data-backed UX strategy can solve real business problems and drive conversion.
