Overview
Dexco (formerly Duratex) is a publicly traded Brazilian manufacturer — the largest producer of wood panels, bathroom fixtures, and metals in the Southern Hemisphere, listed on B3 with roughly USD 1.2B in annual revenue. For most of its history it sold through distributors, retailers, and specification channels. It did not sell to people.
In 2021 the company committed to building its own commerce platform: a direct digital channel consolidating a catalogue that had only ever existed as distributor price lists and specification documents. I joined as Senior UX Design Consultant to lead the design side of that transition — owning UX strategy across Discovery and Delivery, coordinating two independent design teams (one internal, one outsourced), and working across the App 1 Click, Workspace, and Dexco Marketplace workstreams.
The core design problem was not building an e-commerce site. It was that a manufacturer's product data model and a shopper's decision model are fundamentally different structures, and the platform had to reconcile them for two audiences who shop nothing alike.
Business Context
Dexco's move to direct digital commerce was a channel strategy decision with a design dependency: the platform could only work if the company's existing product knowledge could be restructured around how buyers actually decide.
Three business realities shaped the work:
The catalogue was organized for manufacturing, not for buying. Tens of thousands of SKUs existed in structures reflecting production lines, material categories, and distributor commercial terms. A consumer looking for a bathroom does not think in wood-panel product families. Making that catalogue navigable was a taxonomy problem before it was an interface problem.
Two audiences, opposite journeys. Architects and specifiers arrive with technical requirements, compare on specification, and buy in project quantities. Consumers arrive with a room and a budget, compare on appearance and price, and need guidance. A single platform had to serve both without making either feel it was built for the other.
Channel sensitivity in a public company. Selling direct means engaging with existing distribution relationships, and operating inside a listed company meant that roadmap information, revenue projections, and launch timing carried disclosure implications. Design decisions had to be made and communicated inside those constraints.
Why This Was Hard
There was no MVP, no existing digital customer base, no analytics. Personas and journeys had to be built from marketing research, customer success knowledge, and the tacit expertise of Dexco's in-house architects — reconstructing user understanding from proxies rather than observing it directly.
An internal team and an outsourced team worked parallel streams, remote-first, on a shared platform. Without a governing structure, two teams solving adjacent problems independently produce two products wearing one brand.
At tens of thousands of SKUs, navigation, filtering, and taxonomy stop being layout decisions and become the product's core usability determinant. A wrong category model isn't a design flaw you patch later — it's a structure every subsequent screen inherits.
Building two separate experiences would have doubled cost and fragmented the catalogue. Building one generic experience would have underserved both. The resolution had to live in adaptive pathways through shared infrastructure.
The initiative had executive attention and a fixed launch window, running against real platform constraints. Design worked in rapid prototyping cycles because there was no room for long exploratory phases.
My Role and Ownership
I led UX for the initiative as Senior UX Design Consultant, across strategy, execution, and team coordination:
Directed Discovery and Delivery, translating a commercial channel strategy into user journeys, taxonomy, and interface decisions that engineering could build against.
Coordinated two independent design teams — internal and outsourced — establishing shared frameworks so parallel streams converged rather than diverged.
Owned design across three workstreams: App 1 Click, Workspace, and Dexco Marketplace.
Built the research foundation without a product, running empathy mapping workshops and constructing personas from marketing research, customer success insight, and interviews with in-house architects.
Designed the specifier journey with domain experts. Working directly with Dexco's architects, I defined product recommendation flows reflecting how professionals actually specify materials — not how a generic recommendation engine assumes they browse.
Delegated by designer maturity, assigning work to stretch each designer without risking deliverables — mentorship structured into the delivery plan rather than added alongside it.
Presented and defended design decisions to C-level stakeholders in rapid iteration cycles, aligning design intent with commercial expectations and engineering constraints.
Approach and Key Decisions
Reconstruct the users before designing for them.
With no MVP and no digital customer base, I treated persona development as a research synthesis problem: empathy mapping workshops surfaced the friction of customers accustomed to shopping across multiple stores, while marketing data and customer success knowledge filled gaps direct observation couldn't. The output was explicitly labelled as inference-based, so the team knew which assumptions needed validating post-launch.
Design product taxonomy as the primary UX deliverable.
With tens of thousands of SKUs, how products are categorized, named, and related determines whether search, filtering, browsing, and recommendation work at all. I treated taxonomy as foundational design work rather than content operations — because every downstream interface decision inherits it.
One platform, two pathways.
Rather than splitting into separate specifier and consumer products, I designed differentiated entry points and recommendation logic on shared navigation and checkout infrastructure. Architects get specification-led discovery; consumers get guided, outcome-led journeys. The catalogue, the cart, and the checkout stay unified.
Let domain experts shape the recommendation model.
Dexco's in-house architects knew which products genuinely go together in a real specification. Encoding that professional judgment into recommendation flows produced suggestions that were credible to specifiers — something a purely behavioral model couldn't have delivered with no historical transaction data to learn from.
Govern the two teams with frameworks, not reviews.
Consistency across distributed teams cannot be enforced by approving every screen. I built shared UX frameworks — navigation conventions, product taxonomy rules, checkout flow patterns — so both teams made compatible decisions independently. Governance moved upstream into the structure rather than downstream into review.
Prototype at the speed of the decision cycle.
With C-level stakeholders and a fixed window, rapid prototyping served a specific function: giving executives something concrete to react to early, when redirection was still cheap.
Design System Impact
The frameworks built to coordinate two teams functioned as the platform's shared design infrastructure:
Taxonomy as a system. Category structure, product naming conventions, and attribute models were defined once and applied across both teams' streams — the highest-leverage standardization possible in a catalogue-driven product.
Navigation and checkout conventions established as shared patterns, ensuring users experienced one product regardless of which team built which surface.
Recommendation patterns codified from the architect collaboration, reusable across specifier and consumer contexts with differentiated logic on shared components.
Cross-team governance artifacts that reduced rework across distributed streams — the mechanism by which parallel delivery stayed coherent.
UX practice adoption inside Dexco. The frameworks outlasted the engagement, raising the baseline for how product teams across the organization approached design.
Outcomes
All critical MVP features delivered within the project timeline, enabling Dexco's entry into direct digital commerce.
50,000+ SKUs made navigable at launch through taxonomy, filtering, and navigation design built for catalogue scale.
Design governance process that measurably reduced rework across two distributed teams — the structural fix that made parallel delivery viable.
A commerce initiative supporting substantial annual revenue, positioning design work directly against a headline business transformation.
Two audiences served by one platform, with specifier and consumer pathways sharing infrastructure rather than fragmenting into separate products.
Increased UX adoption across Dexco's product teams, extending the engagement's impact beyond the platform itself.
Key Learnings
When there's no product, research becomes synthesis. Building credible personas from marketing data, customer success knowledge, and expert interviews taught me to be rigorous about labelling what's evidenced versus inferred — and to design validation into the launch plan for everything in the second category.
Taxonomy is the interface. At catalogue scale, category structure determines usability more than any screen does. The most consequential design decisions on this project were made in spreadsheets, not in Figma.
Governance scales better than review. Two teams stayed aligned because they shared frameworks, not because I approved their output. Shared structure is how design consistency survives distributed delivery.
Domain experts hold research you can't run. Dexco's architects had decades of pattern knowledge about how materials get specified together. Treating them as a research source, rather than as stakeholders to inform, materially improved the product.
Public-company constraints are design constraints. Working inside disclosure boundaries shaped what could be shared, when, and with whom — a real constraint on collaboration that had to be designed around rather than complained about.
Closing Statement
Dexco was a case where design's job was translation: from a manufacturer's catalogue into a shopper's mental model, from a wholesale channel strategy into a consumer experience, and from two independent teams into one coherent product. The lasting output wasn't the launch — it was the taxonomy, the frameworks, and the governance model that let a company with no direct-to-customer history start operating like one.