Company
Shufti — global KYC/AML identity verification platform
SHUFTI PRODUCT TRANSFORMATION · GLOBAL EXPERIENCE SYSTEM
Shufti: Local Depth at Global Scale
How I designed a governed country-page system built to support 200+ localised page experiences
- One governed system
- Designed to support 200+ localised pages
- 9 priority markets audited
- Live in Germany, Singapore and Brazil
The 30-Second Version
Foundation · market evidence · governance
Shufti verifies identities across 200+ countries — and every market needs a page that feels locally credible: local regulators, local documents, local language, local data rules. As Senior UI/UX Designer, I designed the experience system that makes that possible without turning every market into a separate product.
- Architected a governed country-page system designed to support 200+ localised page experiences from one shared structure.
- Audited nine priority markets (USA, UK, India, Brazil, Germany, UAE, Australia, Canada, Singapore) and converted recurring issues into system rules.
- Defined a three-layer grammar — a stable experience foundation, a variable market-evidence layer, and governance gates for review, publication and activation.
- Live, inspectable implementations across Germany, Singapore and Brazil, including a Brazilian-Portuguese language variant.
The core design idea: standardise the way the experience works; localise the evidence that makes it credible.
Project Overview
Year
2026
Role
Senior UI/UX Designer
Primary focus
Global country-page architecture, localisation rules, governance and activation readiness
Research approach
Nine-market audit, workflow mapping, four-competitor benchmark, live-page inspection, localisation stress testing
Team
Cross-functional programme — marketing, content, SEO, compliance, legal, product, engineering, sales and leadership; I owned the experience architecture and governance design
Duration
2026 · April 2026 audit snapshot; system introduced across the ongoing country-page programme
Tools
Figma (system, QA) · Figma AI · Claude + reusable Claude Skills
Related case
Designing Confidence at Scale — the enterprise decision journey
The Problem Was Not Translation
A country page in identity verification has to do much more than swap languages. It must explain the local regulatory environment, show which identity documents and verification methods matter, clarify business and individual checks, address data and deployment concerns, and give a buyer a credible path to evaluation. A page can be perfectly translated and still fail to feel locally useful.
The programme was expanding across several kinds of variation at once. Some markets needed dense regulatory explanation; others emphasised local documents, eID methods, name handling, data residency or business identifiers. Languages stretched headings unpredictably. A legal update could invalidate a polished page overnight. And marketing and sales still needed one consistent conversion path through it all.
Standardise the way the experience works. Localise the evidence that makes it credible.
What the Nine-Market Audit Revealed
I audited nine priority-market pages — USA, UK, India, Brazil, Germany, UAE, Australia, Canada and Singapore — in the April 2026 operating snapshot. There was no single universal problem. One page needed a legal-content update after a regulatory change. One still used an older template. Two required language-consistency decisions. Four contained repeated modules that needed a tighter flow. CTA hierarchy and tracking readiness varied across the whole set.
That changed the project. The goal was no longer to produce more country pages — it was to build a system that could tell whether any page was structurally coherent, locally credible, legally reviewable, technically ready and safe to activate.
From Audit Issues to System Rules
The audit became useful only when each recurring issue turned into a rule with inspectable evidence:
Audit issueSystem rule
Template variation
A stable foundation with version and status control; shared hero, grid and CTA structure across live pages.
Legal freshness risk
Source ownership, status tracking and a legal-review gate before publication.
Language inconsistency
Translation and language-review states; German and Brazilian-Portuguese treatments operate inside the same layout.
Repeated modules
Core, conditional and optional module rules, so evidence can be added or removed without breaking the page spine.
CTA and tracking variation
A shared CTA hierarchy plus an activation gate separating publication from promotion.
The Central Design Tension
A single rigid template would be easy to build but would flatten the differences that make a market page useful. Designing every country independently would protect local expression but multiply layout decisions, QA risk and long-term drift. I compared three system directions — fixed template, fully custom, and a governed modular grammar — and selected the grammar because it gives every discipline the control it actually needs: marketing varies the narrative, compliance reviews claims inside predictable modules, engineering builds from stable components, SEO manages intent without rebuilding structure, and design protects hierarchy and conversion consistency.
The Three-Layer Experience Grammar
- 1
The experience foundation
How every page behaves before local content exists: grid, typography, spacing, navigation, CTA hierarchy, cards, forms, trust treatment and responsive rules.
- 2
The market-evidence layer
Everything that must change per market: regulators and authorities, accepted documents, languages, local identity methods, statistics, relevant products and industries, data controls, AML sources and FAQs. Modules are mandatory, conditional or optional — but their visual behaviour is always defined.
- 3
The governance and activation layer
How a page moves from content to live traffic: AI-assisted drafting, manual fact-checking, translation and translation review, asset creation, Figma QA, lead review, development — then activation gates for QA, working CTA, validated tracking and a connected offer before promotion.
The third layer matters because a page can be visually complete while still legally stale, incorrectly translated, weakly tracked or disconnected from a useful next step. The system deliberately separates design completion, publication readiness and traffic activation.
The grammar itself was iterated, not declared: module rules were revised after German and Portuguese content stress-tested wrapping, density and responsive behaviour, and review rounds with compliance and SEO reshaped which modules were mandatory, conditional or optional. Everything in the system was treated as a prototype until real market content survived it.
Designing the Page Narrative
The page anatomy is a grammar, not a fixed ten-section template. A core spine establishes the market promise, local verification evidence, regulatory context, product relevance and conversion. Conditional modules appear only when KYB and beneficial ownership, data residency, industry applications, AML-source depth or operational statistics materially matter in that market. Optional modules carry extended FAQs, guides, extra documents and related content.
Ordering rules keep it coherent: regulation moves earlier when it drives the evaluation; data controls appear when sovereignty or hosting matters; product modules follow the market problem they resolve; and the primary CTA keeps consistent hierarchy no matter how the evidence changes.
The System in Use
Regulation and identity-method depth
Visible alignment with the German Money Laundering Act, BaFin supervision and EU obligations; Personalausweis, electronic residence permit, eID/NFC considerations, German name variants and EU data controls — all without overwhelming the conversion path.
Identity, business and language context
MAS supervision, STRO reporting, NRIC and FIN credentials, UEN-based business identification and PDPA-aligned data considerations — a different evidence profile on the same foundation.
Domestic credentials and Portuguese localisation
COAF, Central Bank supervision and LGPD alongside RG, CIN and CPF credentials. The Portuguese variant proves that copy length, terminology and evidence can change while hierarchy and conversion stay recognisable.
Designing for Language and Content Pressure
Localisation exposes failure states an English desktop page never shows. Long German regulatory labels and Portuguese copy change wrapping and density. Some markets need more document cards, longer source lists or deeper FAQs. So I specified behaviour, not fixed screenshots: headings wrap without shrinking CTA prominence; cards grow vertically while keeping internal spacing; optional modules collapse when evidence is missing; long source lists use a controlled reading pattern; and mobile reorders content by decision priority instead of squeezing desktop columns.
Governance: A Concrete Leadership Decision
The most important leadership decision was separating local freedom from structural freedom. Marketing needed flexibility, engineering needed predictable modules, compliance needed fixed review points, SEO needed clear ownership. Treating every request as a new page design kept those needs in permanent conflict — so I converted the disagreement into rules: the foundation cannot change without a system decision; market modules change when evidence justifies it; publication cannot proceed while required review or activation checks are missing. Teams got one shared rule for resolving the same trade-off in every future market.
AI as Controlled Production Support
AI supported the workflow but never owned the truth of a page. Claude and GPT helped with first-pass content, translation drafts and visual concepts; the process still required manual fact-checking, translation review, source validation, Figma QA and human approval. In regulatory content, a fluent sentence can still be wrong for the country — so the design contribution was defining where AI helps, where human judgement is mandatory, and which status a page may hold while evidence is incomplete.
Impact
Architected a system designed to support 200+ localised page experiences — three-layer grammar, module taxonomy, responsive rules and publication logic.
Audited nine priority markets and defined system rules for recurring structural, content, language, CTA and readiness issues.
Shipped inspectable implementations across Germany, Singapore and Brazil, including a Brazilian-Portuguese variant on the same underlying structure.
Introduced governance and activation states connecting design, content, legal, SEO and engineering handoffs into one visible status model.
Created reusable market modules and publication checks so teams can adapt pages without redefining the structure. Deployment speed remains unmeasured.
What I Would Measure Next
Connect the system to page-level performance: qualified demo requests and pipeline by country-page first touch, CTA success, form completion, component reuse, localisation defects, stale legal content, and the time needed to update a market after a regulatory change. Then compare pages activated through the full readiness gate against pages published before validation. That separates three questions the current evidence cannot yet answer: did the system reduce production effort, did pages keep accuracy over time, and did the experiences contribute to acquisition.
Reflection
A global experience does not scale by making every market look the same. It scales by making variation predictable, reviewable and safe to implement. The most reusable design decision was not a card, a template or a hero layout — it was the separation of foundation, evidence and governance, which let one interaction language give every market enough room to be genuinely local.
What I would do differently: attach analytics ownership to the activation gate from the very first market, so that every page entering the system started generating the performance evidence the current case cannot yet claim.
Global consistency should protect the experience. Local depth should earn the variation.
