Four legacy systems,
one placement platform

The strategic problem

I had spent seven years in Howden’s internal design team, designing several of the 1.0 placement platforms. Each product line ran on its own stack, and the cost compounded in four directions at once.

Duplicated maintenance across multiple tech stacks. Launching a new product line meant building from scratch. Inconsistent data capture made analytics unreliable. And brokers had to learn a different interface for every product they touched.

My role

I joined as lead product designer and evolved into product owner. That transition happened because the problems required someone who could hold the user experience and the delivery reality in the same frame.

In practice: owning the roadmap and the prioritisation calls, leading a 17-person delivery team through standups and sprint planning, running weekly research with brokers and underwriters, managing a £1.5M annual budget, and partnering with North — the agency behind the Tate Modern and Co-op identities — on Howden’s digital design language.

The Outlook mental model

Enterprise software fails when it asks users to abandon mental models they have spent years building. From the first round of interviews one thing was obvious: brokers live in Outlook.

So rather than imposing a novel paradigm, I leaned into what they already knew. A pane-based layout mirroring Outlook’s list-plus-detail pattern. Clear hierarchy for submissions, placements and market responses. Interaction patterns that felt immediately familiar.

We tested with groups of up to ten brokers. Onboarding friction was close to zero — people understood how to navigate in minutes rather than days.

The configuration tool nobody asked for

As we onboarded more product lines, a bottleneck emerged. Product configuration was being done by hand in JSON. Early configs ran to a few hundred lines; they ballooned into the thousands, with field IDs reused across subsystems.

This was the constraint on our ability to scale, and it was invisible on any roadmap. Rather than asking for a dedicated engineering initiative, I prototyped the tool myself: automated ID generation and mapping across subsystems, product logic exposed visually instead of buried in raw JSON, validation before deployment, and a structure legible to people who do not read code.

Configuration timelines dropped by roughly a month per product. We have since launched five more products through it.

Export over dashboard

Midway through the build, stakeholders asked for a dashboard. Charts, KPIs, visual reporting. I pushed back.

Having led Howden’s dashboard unification effort, I knew how dashboards actually get used here: people export to Excel, rebuild the view they need, and circulate a spreadsheet. So I proposed investing in export instead — clean column selection and filtering, formatting aligned to Excel workflows, and reliable performance on large datasets.

It met the real need at a fraction of the cost, and it scaled automatically as we added products. No dashboard redesign required.

Design system and brand

With 2.5 designers on the team, I led a system that standardised the components the platform actually runs on — tables, forms, pricing panels, filters — and defined typography, spacing and interaction patterns for consistency and accessibility, aligned to the brand direction we developed with North.

It has since been adopted across other Howden applications, which multiplied the original investment well beyond xTrade.