Delivering the next generation of credit intelligence
Atrium Insights is an enterprise platform used by organisations, brokers, and internal teams to monitor credit exposure, policy performance, and portfolio trends. More than 3,500 users rely on it every day.
Insights 2.0 was a modernisation effort, bringing together a new technology stack, refreshed designs, and smarter capabilities. Many users had relied on the previous platform for over a decade. The challenge wasn't to change their workflows, but to make the experience faster, simpler, and more intuitive.
Users relied on a legacy platform, but expected a faster, more intuitive experience.
Research didn't stop after discovery. We introduced a lightweight feedback loop that ran alongside delivery, allowing us to validate ideas without slowing development.
Every quarter, we gathered feedback through usability and multivariate testing with internal and external users, in-platform surveys, coordinator check-ins, and interviews.
Findings were synthesised, prioritised with the Product Owner, business lead, and engineering team, then fed into upcoming releases. This ensured roadmap decisions were informed by evidence while committed work continued uninterrupted.
By the end of the project, this had become part of how the team planned and prioritised future work.
Diagram: Process of prioritising future work.
Using feedback to influence the roadmap
In Q1 2026 roadmap, the original plan included the Policy Results Dashboard and Speed of Service improvements. During a coordinator survey, Policy Performance and Declarations consistently ranked as higher priorities, so the roadmap was updated.
In Q3 2025, with 100+ user lane comments and findings from interviews. Common pain points around policy group creation, table filtering, and overall performance were prioritised and addressed in the following quarter.
Before major features were released, we validated them through usability testing, surveys, and multivariate testing. When feedback highlighted confusion or uncovered new issues, we investigated further before moving ahead.
Figures 2 and 3: Original plan and proposed new plan for Q1 (based on user feedback)
Making powerful decisions easier to use
Legacy products have more life in them than people think. You just have to keep listening.
Searching and filtering large portfolio tables is one of the most common tasks in Insights, making it a key area to improve. Search was originally handled through a 'Filter Table' button, which revealed search fields above every column. On larger tables, this meant more than 20 visible search inputs. While the functionality was powerful, the interface felt cluttered and difficult to scan.
We tested three different approaches through multivariate testing, including a centralised search panel, grouped filters, and a simplified version of the existing interaction. Six out of ten users preferred keeping the familiar behaviour, provided it was easier to understand.
Based on this, we retained the existing interaction but renamed 'Filter Table' to 'Search', making its purpose clearer without changing a workflow users already knew.
Since the number of columns was driven by business requirements, reducing them wasn't an option. Instead, we focused on giving users more control by allowing them to resize, reorder, pin, and show or hide columns.
Building on this, users could customise a table once and reuse that setup. Working with engineering, we implemented this using NUQs, allowing table configurations to be stored in the URL so they could be shared, revisited, and applied across other tools. 73% of participants said this would save them time.
A complete redesign of the filtering experience remains on the long-term roadmap. For this release, improving discoverability and flexibility delivered the greatest value.
Figures 4 and 5: Old Insights, user can only filter the table with a column search. Figure 5 has redesigned insights, a dedicated search toggle that enables search for each column; right click enables a menu to customise display.
Enhancing without disrupting familiar workflows
The old Insights colour scheme -yellow, ochre and grey, wasn't brand-aligned or accessible. However, it was familiar to people who had logged in every day for years.
Rather than redesign everything, we focused on changing only what needed changing. The visual language was rebuilt from scratch to be accessible, brand-aligned and modern, while the overall layout remained familiar. Buttons that already worked stayed where users expected them. Flows that caused confusion were redesigned.
Enterprise users build strong muscle memory over time. Keeping the structure familiar reduced the amount users had to relearn while still delivering a noticeably better experience.
Across more than 100 user lane comments and A/B testing, users consistently described the redesign as easier to digest, more modern, and "feeling Atradius."
Figures 6 and 7 show the buyers tool.
Evolving a trusted platform to meet future user needs
One of the goals of Insights 2.0 was to help users understand portfolio data faster, without asking them to interpret dense visualisations. We redesigned several key data visualisations, including simplified chart representations, a toggle for 'top 5 buyers', a 'best trades last quarter' summary view, etc. In 5/6 show and tell sessions, users found them highly useful.
The Buyers tool became the clearest example. Users described the previous version as dense and difficult to interpret. I retained the familiar layout but made the insight more obvious by replacing colour-coded buyer ratings with explicit statuses like Improved, Deteriorated, and Unchanged. Clearer legends, optional gridlines, and a Top 5 Buyers view further reduced the effort needed to understand the data.
Although these concepts tested well, they weren't included in the first release because the underlying data platform was still being rebuilt. Rather than introducing richer visualisations on an unstable foundation, the team prioritised platform stability first. The concepts remain validated, supported by the Product Owner, and are planned for a future release
Figures 8 and 9 show the dashboard.
Conclusion
Research became a product capability, not a project phase
Rather than conducting research once and building against static insights, Atradius embedded discovery into every quarter. This created a continuous feedback loop, allowing the platform to evolve alongside changing user needs and ensuring decisions were based on current evidence, not assumptions.
For a legacy product, success wasn't just about building new features. It was about maintaining a clear vision, validating priorities with users, and creating a roadmap that balanced immediate improvements with longer-term platform ambitions.