BINGO Admin Console
The back office console that compliance officers had routed around with personal spreadsheets, rebuilt so a one minute review takes one minute. Task completion rose from 42% to 92% and the SUS score climbed from 28 to 81.
At a glance
Task completion rose from 42% to 92% and the average search fell from 4:28 to 1:25, measured with the identical benchmark protocol before and after.
The SUS score climbed from 28 to 81 — a failing grade to an A.
I led the console redesign and mentored two new UX designers, Kezia and Angelica, who finished the project running their own reviews.
The information architecture was rebuilt from card sorting with compliance officers and tree-tested before any screens were drawn.
42% to 92%
Task completion
4:28 to 1:25
Average search time
28 to 81
SUS score
The problem
Back-office tools set the pace a regulated business can move at, and BINGO's admin console was the bottleneck. KYC compliance officers lost hours each day to it. Search matched exact prefixes only, so a misremembered client name returned nothing. Filters were three separate dropdowns that couldn't stack. Column headers used internal system codes nobody outside engineering understood. Task completion sat at 42%, and the average search took 4 minutes 28 seconds, for work the business needed done in one. The brief was operational efficiency at enterprise scale: make a data-dense system legible to the people who run it all day, and take the cognitive overhead out of every review.
Strategic alignment
The situation
The challenge
BINGO is a Southeast Asian fintech platform with regulated KYC requirements, and its admin console set the pace the business could move at. Compliance officers reviewed thousands of submitted client documents in a system that had grown ad hoc for two years with no design review. Search matched exact prefixes only, so a misremembered name returned nothing. Filters were three separate dropdowns that could not stack. Column headers used internal system codes nobody outside engineering understood.
The benchmark made the cost concrete: 42% task completion, a 60% error rate, an average search of 4 minutes 28 seconds for a one minute job, and a SUS score of 28. Officers had quietly built personal spreadsheets to route around the system entirely.
My role and contribution
I was the senior UX designer leading the console redesign and the design team behind it. Two new UX designers, Kezia and Angelica, joined the project under my direction and mentoring, alongside two frontend engineers and one backend engineer, over six months. I set the research and design direction and stayed hands on: the interviews with 7 compliance officers, the card sorting and tree testing that rebuilt the information architecture, the interaction design for search, filtering and export, the copy review with the compliance team, and the benchmark and retest programme that proved the result. As the project matured, Kezia and Angelica took growing ownership of screens and usability audits.
The brief was operational efficiency at enterprise scale: make a data dense system legible to the people who run it all day, and take the cognitive overhead out of every review.
Key objectives
Make search trustworthy
Every officer's first complaint was that they could not trust search. Partial names and IC numbers had to return results, every time.
Make filters visible and stackable
Replace three separate dropdowns with filters an officer can see, combine and clear in one click, without opening a panel.
Speak the officers' language
Rename every label to match how officers describe their work, driven by card sorting rather than internal product categories.
Prove it with the same ruler
Re-run the identical benchmark protocol after launch, so the before and after numbers are directly comparable.
Studying the category
I interviewed 7 compliance officers over two weeks, ran card sorting with 5 of them, and tree tested the proposed structure before touching any screens. All seven said the same thing: they could not trust the search. Card sorting showed the labels did not match how they thought about their work. They grouped by client status and date, while the console grouped by internal product categories that meant nothing to them.
Systems thinking
The console was restructured from the information architecture up. Card sorting revealed how officers actually group their work, tree testing validated the new structure before any interface was drawn, and the primary task flow was mapped as one clear path with no dead ends and no backtracking.
An IA rebuilt from card sorting
Five officers sorted the console's content by client status and date. The internal product categories meant nothing to them, so the new structure follows the mental model of the people who do the work.
Tree tested before any screens
The proposed architecture was validated with tree testing before high fidelity work began, so structural mistakes were caught while they cost hours instead of sprints.
One path through the task
Search, filter, review, export, mapped as a single flow. Every screen answers one question: what does the officer need next to finish this review?
Process archive — the full flows & structure, 2 artefacts for the deep read
Define & structure
Open card sort with 5 compliance officers. Each officer grouped 40 content cards independently. Results showed consistent grouping by client status and recency. Nobody grouped by BINGO's product category taxonomy.
Fuzzy search with edit-distance matching. Autocomplete suggestions surface at 2 characters. Handles partial names, IC number fragments, and common misspellings.
Information architecture
The solution
Three changes drove the result, each answering a specific failure the research had isolated.
Search that forgives
Fuzzy, edit distance matching replaced exact prefix search, so partial names and IC numbers still return results. Filter chips replaced the stacked dropdowns: they stay visible, combine freely and clear in one click, so officers always see what is active.
- Autocomplete appears from the second character, closing most searches before they are fully typed.
- Edit distance matching tolerates the misrememberings that used to return nothing.
- Active filters read at a glance instead of hiding inside three dropdown panels.
Fuzzy search and stackable filter chips, so officers find what they need in one action.
Labels in the officers' language
Every ambiguous column header was renamed to plain language, and contextual tooltips explain the remaining domain terms on hover. The copy went through review with the compliance team, so the explanations match how officers describe the fields to each other.
- Internal system codes were replaced with names from the officers' own card sorting vocabulary.
- Tooltips surface detail on demand instead of crowding the table.
- The compliance team reviewed every line, keeping the language accurate as well as plain.
Human readable data labels, with tooltips surfacing detail on demand.
Export where the work ends
Every review ends in an export, yet the action sat three levels deep in menus. It became a persistent button, always visible and always reachable, so the last step of the task is also the easiest.
- One click from any state of the table, honouring the active filters.
- No navigation away from the review in progress.
- The most frequent end of task action is now the most available one.
Persistent one click export, moved up from three levels deep.
Before & after
All screens
Outcomes
Where it landed
The rebuilt console shipped to production and the retest used the identical protocol as the benchmark: task completion rose from 42% to 92%, the average search fell from 4 minutes 28 seconds to 1 minute 25 seconds, and the SUS score climbed from 28 to 81, a failing grade to an A. The clearest signal came from behaviour rather than surveys. The personal spreadsheets officers had built to route around the old system quietly disappeared.
What the work taught me
The console moved every number it was measured on, but the sharper lessons were about where the answers had been hiding all along.
Workarounds are a map
The officers' personal spreadsheets were the real research finding. When people build their own tools around your product, they are drawing you a map of everything it fails to do, and the fastest wins came from reading that map rather than inventing from scratch.
Language is design work
Half the console's confusion came from labels written by the system's builders instead of its users. Renaming things is unglamorous, but it moved the SUS score as much as any interaction change.
Measure with the same ruler
Running the identical benchmark protocol before and after is what made the result credible. A 50 point completion lift is easy to claim and hard to argue with when the method never changed.
Mentoring is a multiplier
Kezia and Angelica joined as new designers and finished the project running their own reviews. Bringing them into real research sessions early, then handing them screens with clear guardrails, grew the team's output faster than any process change, and it is the part of this project I am proudest of.
