Redesigning how enterprise sales teams navigate product complexity — transforming a fragmented catalog into a decision-support system that serves client needs, not internal org charts.
"The brief was wrong. That turned out to be the most important finding of the entire project."
A financial technology company builds and sells banking products — credit solutions, treasury tools, lending instruments — and demos them directly to enterprise banks. Their internal platform was supposed to help sales reps find the right product and build a compelling pitch for each client meeting. The brief was straightforward: better search, cleaner navigation, more modern UI.
The research told a different story. Sales reps weren't lost in a catalog. They were making high-stakes decisions under pressure — about which product to lead with, how to position it for a specific bank's risk appetite, and how to assemble a coherent pitch from 8+ disconnected internal sources during a live demo. The platform had been designed to store product information. It needed to be redesigned to support decisions in the room.
That single reframe changed the IA, the product strategy, the success metrics, and ultimately the business outcome. The platform that shipped was categorically different from the platform that was specified — and it was better for it.
Understanding why was the most important design work on the project.
"Build a better product marketplace — cleaner search, better navigation, modern UI."
A catalog of banking products that sales reps could browse, filter, and present to clients. The original brief framed this as an information architecture problem: organise the products better and the reps will find what they need.
"Sales reps aren't lost in a catalog. They're making decisions under pressure."
Shadow sessions revealed the actual workflow: reps were juggling 8–12 tabs, cross-referencing PDFs, retyping product details from memory, and building client pitches from scratch for every meeting — not because they couldn't find products, but because nothing connected the products to client situations.
The research dismantled the catalog assumption before a single wireframe was drawn.
The 8–12 tab reality. Reps weren't struggling to find products — they were struggling to connect products to client situations. The catalog was the wrong starting point. Client situation was the right one.
Three completely different mental models for the same platform. Sales needed situational guidance. Product needed structured presentation. Compliance needed controlled language. One IA had to serve all three — differently.
Navigation organised around internal product taxonomy, not client situations. No way to compare products in context. Compliance language buried or absent. Every critical workflow required leaving the platform.
New reps took 3+ weeks to reach baseline competency — almost entirely spent learning which products applied to which situations. The knowledge existed in senior reps' heads, not in the platform. Institutional knowledge was invisible and non-transferable.
Each move was a deliberate judgment call — not a feature, a strategic decision about what the platform was for.
Challenged the catalog brief in Week 1 before any UX work began. Shadow sessions confirmed the real problem was situational guidance, not information access.
3 weeks of sales floor observation before touching Figma. The 8-tab reality, the cheat sheets, the institutional knowledge gap — all invisible from the brief.
Replaced product-category navigation with a guided discovery flow starting from client situation. "I'm talking to a mid-market CFO about working capital" — then the platform filters.
Instead of compliance as a legal afterthought, built it into the product content model. Every product entry included approved language, required disclosures, and client-appropriate framing — locked and maintained by product teams.
Built a content governance model into the IA — product teams own their entries, the platform enforces structure. Designed for the platform to stay accurate without a dedicated team to maintain it.
The architecture was built from the sales rep's decision context outward — not from the product inventory inward.
"The goal was to build a system that sales reps could trust completely — not one they needed to interpret."
Five key screens from the final platform — from initial discovery through the complete client-facing experience.
Not design preferences — the specific calls that changed direction, resolved deadlocks, and determined what the platform became.
The first round of designs was excellent at navigating the existing catalog — exactly what the brief asked for. The shadow sessions had already shown the brief was wrong. I killed the work in Week 3 and reframed from situation-based discovery. The stakeholders needed a scenario to feel the problem — a rep ten minutes before a client meeting, still building their pitch from scratch across three tools, no single view of what to recommend or why. That scenario got the room.
Legal wanted compliance language surfaced as alerts after rep selection. Product wanted it hidden from the sales view entirely. I proposed making it structural — built into the content model as a required field, maintained by product teams, invisible in client-facing views but always present in the system. Both sides accepted because neither was being overruled — the design resolved the conflict at the architecture level.
With 72 hours to launch, testing with new hires revealed the guided discovery flow was fast but fragile — reps who didn't fit the defined client profiles hit dead ends. I held the launch, restructured the edge case paths, and ran two additional testing sessions. Three weeks delay. The alternative was launching a platform that would be abandoned by 30% of the user base on day one.
The 40% demo prep reduction was self-reported in follow-up interviews. I should have designed the analytics instrumentation plan before launch — so the platform generated its own evidence from day one. Behavioral data would have been more defensible than survey data, and more useful for the next iteration cycle.
The content governance model worked — product teams maintained it. What I underestimated was how long it would take for compliance language to be properly populated across all products. We launched with ~60% coverage. The full value wasn't realised until month 3. I'd build that migration timeline into the project scope from the start.
When I killed iteration one and proposed the situation-first model, the alignment happened in a room. Six months later, new stakeholders joining the project didn't understand why the platform was designed the way it was. I'd write a decision document that travels with the platform — capturing the reasoning that produced the architecture, so it doesn't have to be re-explained every time someone new joins.