Redesigning How a $2 Trillion Industry Buys Ocean Freight

The Challenge

Ocean freight runs on annual fixed-price contracts that take months to negotiate. When the market moves, the contracts stop working, and the supply chain breaks.

Index-linked contracts solve this by tying rates to a market index, so pricing moves with the market instead of being locked in at the wrong moment. The mechanism works. The problem is that it's complex, unfamiliar to most of the industry, and built on commercial judgement calls that customers need to feel confident making.

My job was to design a product that made those judgement calls navigable rather than intimidating: the first index-linked contract simulator in the world, built to turn indexing from a niche capability into something customers across the market would adopt.

Research & Discovery

I led discovery with the people closest to the problem: customer success managers, who spend the most time explaining indexing, and freight forwarders, who use the Simulator to validate the approach with their own customers. We also analysed how existing index-linked contracts in market were configured to understand which parameters customers were really using.

I ran user-testing sessions directly with shippers, including Walmart and Philips, watching how their procurement teams reasoned about indexed pricing and pinpointing exactly where the mechanics lost them.

One insight kept surfacing: customers trusted Xeneta's data, but not always the mechanics behind it. They believed the benchmarks were accurate. They struggled to understand how an indexed price would actually behave over time, what triggered a rate adjustment, or how to defend the mechanism to their own stakeholders.

That reframed the work. Transparency mattered more than automation. The product needed to make complex pricing logic explainable, not hide it.

From research to product strategy

The workshops surfaced a more fundamental question than how to design any single screen. Customers were entering the world of index-linked contracts from completely different starting points. Some wanted to test the idea before committing. Some had already signed contracts externally and needed somewhere to operationalise them. Some were already running indexed contracts using Xeneta data and needed to monitor them.

I synthesised the research into a product architecture that mapped these journeys end to end: who the personas were, where they came in, and how the surfaces needed to connect. This became the artefact I used to align product, commercial, and engineering on what we were actually building and in what order.

Two decisions came out of it. The first was to launch a free, standalone simulator as the front door. Letting customers backtest the idea without commitment was the lowest-friction way to build understanding, and the simulator could prove value before any contract was on the table. The second was to design the simulator and the live contract manager as one connected system, so the trust built in simulation could carry into operation rather than being lost at handover.

The simulator is the surface this case study focuses on, and the one that won the award. It's also the surface that opened the door to everything else.

Prototyping in code, not slides

I built functional prototypes in code rather than relying on static mockups for stakeholder and customer reviews. It changed how the work landed.

Customers could interact with the work in sessions instead of imagining it, which produced sharper feedback. Engineering had something concrete to align against, which surfaced technical realities earlier and prevented scope from drifting late. Several decisions, including which features to defer, came out of conversations that wouldn't have happened over a Figma file.

Listening at scale: an AI insights engine

V1 shipped and won awards, but it was deliberately simple: a single-lane simulator. To decide what V2 should be, I didn't want to wait on a handful of research sessions. I wanted to hear every customer who had ever talked about the Simulator. So I built the tool to do it.

Xeneta's Customer Success and Account Executive teams record their customer calls in Gong. I built a pipeline that pulls the transcripts of those calls, finds every moment indexing comes up, and feeds each one through the Anthropic API. Claude extracts the new feature ideas and points of confusion, then classifies each by theme, by stance - curious, sceptical, confused - and by account type, with a link straight back to the exact moment in the call.

Across 700+ calls that mentioned the Simulator, that turned anecdote into evidence. Instead of waiting for the next research round, the design team had a continuous, queryable voice-of-customer feed: what customers actually asked for, themed, ranked, and traceable to its source.

This is the part I'm most proud of. Faced with a strategy question, I built my own research infrastructure to answer it - a designer using AI not to generate screens, but to listen at a scale no round of interviews could match.

V2: from simulator to portfolio platform

The insights pointed one way. Customers didn't think in single lanes; they bought freight across whole portfolios, and the question they kept circling back to was how to protect a budget when the market moved. V2 answered both, with a multi-lane portfolio simulator and a Hedging Explorer.

None of it was change for its own sake. Each move answered something V1 - or the research around it - had surfaced.

Multi-lane portfolio simulation

Instead of modelling one corridor at a time, customers can now build an entire contract book - adding trade lanes by origin, destination, equipment and volume, or importing them straight from a watchlist or CSV - and apply one set of indexing rules across all of them. The simulator backtests the whole portfolio against five years of real Xeneta data, including the COVID and Red Sea disruptions, so a procurement lead can see how their book would actually have behaved before signing anything.

The Hedging Explorer

Modelling a contract tells you what your pricing would have done. It doesn't protect you from it. The Hedging Explorer adds that next step: futures and derivatives guidance that lets a customer hedge a contract as well as simulate it.

Research had flagged futures as powerful but risky to surface broadly - there is a line between guidance and financial advice that a data company can't cross. So the tool leads with education. It establishes the customer's position (buying or selling freight), explains hedging in plain English, checks which lanes are actually hedgeable against the listed Euronext futures corridors, then lets the user dial in how much of their volume to cover with a single slider, surfacing the average rate they'd lock in and how much exposure they'd remove. It is explicit throughout that it is guidance, not financial advice.

Branding Xeneta Indexing

For indexing to be taken seriously as a strategic bet, it had to look like a product line, not a feature buried inside the platform. Alongside the product work, I shaped the visual identity for Xeneta Indexing: a consistent look across the simulator, the marketing surface, and the academy content that sat around it.

It had to do two things at once. It needed to feel distinct enough to stand on its own, while sitting comfortably inside Xeneta's parent brand. That meant working within the existing guidelines on type and colour, then giving indexing its own accent and language so a customer could tell, at a glance, that this was a new class of product.

The result gave the whole programme a single, coherent face - the same identity carrying customers from the marketing page, into the free simulator, and on into their live contracts.

What I learned

Trust is a UX problem. In enterprise products, especially financial ones, transparency is more valuable than simplification. Customers making commercial decisions don't want a black box. They want visibility, confidence, and the ability to defend the result to someone else.

The best product research happens where the product isn't. Watching how a sales team actually sells indexing taught me more about how the simulator needed to feel than any usability test could have. Domain authority belongs with the user, and the product's job is to bring the data, the context, and the patience to let them work it out.

AI can be research infrastructure, not just an output tool. The most valuable thing I built for V2 wasn't a screen - it was the pipeline that let me hear 700+ customer conversations at once. Using AI to scale listening, rather than to generate artefacts, changed what evidence the design team could work from. I think that's where a lot of the craft of senior design is heading.