ARIC Risk Hub

Work for Featurespace

Existing NDA & financial security requirements restrict available visuals

One of the Featurespace Design team's main roles is guiding the continual evolution of ARIC Risk Hub

ARIC is a web-based B2B hub that uses ML to enable fraud prevention for customers such as HSBC & Walmart. Used by financial experts, who conduct AI optimized data analysis in ARIC, enabling real time consumer protection.

As a company, Featurespace relentlessly listens to customers through the many customer facing teams that manage ARIC’s technical implementation, a process that is unique to each customers existing operational flow.

But instead of looking at our customer’s tech stacks, I focus on researching the users within our customers , who need to use ARIC everyday to prevent fraud.

This combination of customer feedback sources drives the UX of ARIC forward in two ways:

Finding better ways to do what users currently need to do
Building new ways for users to do what they want to do

Rethinking data filtering

Type 1 Project
Primary Designer
Dec 2021

ARIC allows analysts to review individual fraud cases, inside of each they are presented with several sets of data. The analysts that use ARIC rely on seeing connections between hundreds of these unique data points.

This experience needs to help analysts be fast & accurate

In this project, complaints of the existing filtering workflow were relayed to myself from Featurespace’s customer success team. I was the main designer challenged with investigating and solving the cause of these frustrations.

To investigate, I led several customer interviews till thematic saturation was reached.

The common thread between all users saw that the filtering experience felt disjointed and cluttered. On investigation into its prior development I found it had been a victim of infamous feature creep. 

Many of the interactions inside the filter were grouped together simply because that was where extra functions were easy to put.

The back and forth workflow interactions involved in controlling real data had never truly been considered.

Building up a comprehensive view of the analysts journey when they review fraud data helped us understand the friction the existing filter design was generating.

These additional friction points were slowing down our users, which results in lower accuracy due to external time limits

From these friction points, four potential journeys were ideated to address them. These were drafted as Hi-Fi interaction prototypes within Figma in order to test with users to see if they improved the speed at which analysts could control realistic data sets.

User feedback & testing sessions were conducted alongside a Featurespace product manager. After consolidating the feedback we found that while we could mitigate many of the friction points the engineering cost of untangling the existing filter would be better served building it anew.

See this related Project

Our findings of user's filtering needs were instead used to fuel a wider project that looked at reinventing ARIC from scratch

Anti-Money Laundering (AML)

Type 2 Project
Primary Designer
Mar 2022

Money laundering is very specialised fraud where governments 'compel' individual companies to responsibly report instances.

While financial fraud costs banks money,  money laundering is technically beneficial for a bank. This is why government’s force companies to care through external fines & criminal charges.

This external scrutiny means customers need to be sure money laundering activities do not get missed. This led Featurespace to identify the need for ‘contributing events’ inside of ARIC.

Okay, that's all good but what actually are 'contributing events'?

Current Experience

Okay, ARIC has highlighted data for us that it is suspicious of, but we don't know how exactly it saw these patterns.

With Contributing events

With 'contributing events' ARIC can now show us the original data that led to its suspicions. This allows analysts to grasp a better picture of the case, reducing doubt and improving decisioning accuracy.

The experience challenge here is how might we enable analysts to better visualise the reasons why ARIC viewed data as fraud?

Several user stories were ideated, looking at how the visulised data would cross & alter between the multiple key zones that analysts have to work in ARIC to complete a case.

Concurrently, the technical feasibility of ARIC's backend machine learning outputs became clouded

To help align the team these storyboarded concepts were altered to incorporate the impacts of several different technical boundaries. In total nine Mid-fi interaction prototypes were built in Figma and used to internally validate the technical requirements for ARIC.

These were then streamlined down into 3 Hi-Fi experiences. In order to gain valuable insight from analysts, extensive use of Figma's interactive component's was made. When deeply layered these delivered highly interactive prototypes in a fraction of the time over dedicated prototyping software such as Axure.

Feedback such as aligning some functionality to user's existing mental models helped improve usability and was implemented. Once the finalised designs had passed accessibility tests conducted with WCAG AA they were written up for handover to the engineers in the team.