Case Studies > Payments

UX Audit & Redesign Direction

Payments Backoffice: Where Money Meets Usability Risk

A heuristic evaluation and practitioner-informed field study across 17 payment screens, from audit findings to redesign direction.

Screens Reviewed

Findings Documented

Payment Flows Redesigned

Practitioner Scenarios

Role: Sole UX Researcher & Designer

Methodology

// Screen review, heuristic evaluation, simulated field research, AI-assisted cross-validation

01 / SCREEN REVIEW

17 production screens walked through, including payment methods, transaction details, withdrawal flows, and the in-progress Lifecycle redesign.

02 / HEURISTIC EVALUATION

Nielsen's 10 heuristics applied with severity scoring (0–4). Key violations found in payment creation and verification flows with compliance exposure.

03 / PRACTITIONER-INFORMED SCENARIOS

3 practitioner-informed scenarios based on observed workflow patterns, including Tier-1 ops, fraud analyst, and compliance. Hypothesis set for live validation.

04 / AI-ASSISTED CROSS-VALIDATION

Ran a parallel exploratory analysis using AI tools against the same screen set, comparing AI-surfaced issues with manual findings to validate coverage and test where automated heuristic detection aligns or diverges from practitioner-grounded research.

Usability Findings// 12 findings across payment flows

Where money meets usability risk

Critical · 1

High · 3

Medium · 7

Low · 1

12 findings ranked by financial exposure and frequency. The critical issue, a Verified badge persisting on garbage bank data, represents live compliance risk.

Verified Badge on Invalid Data

critical

Free-text bank fields accept garbage input even after Verified status, breaking KYC/AML integrity.

Partial Amount Exceeds Approved

high

Partial amount can exceed the originally approved amount with no cap, a logically contradictory state.

Raw JSON Blocks Investigation

high

Transaction remark is an unformatted JSON blob. Agents hand-parse it to find auth codes under chargeback deadlines.

Misleading Withdrawal Action Labels

high

Withdrawal action labels are ambiguous. "Cancel" unclear whether it cancels the request or the agent's action; "Decline" unclear whether it rejects permanently or sends back for review. High-stakes confusion on money flows.

No Bulk Withdrawal Actions

medium

No bulk actions existed. Agents processed one withdrawal at a time, creating repetitive manual work and slowing high-volume shifts significantly.

Transaction ID Not Linked to Details

medium

Transaction IDs in the payments section are not clickable. Agents must navigate to the Transactions tab separately to view details. Breaks contextual flow during withdrawal review.

Icon-Only Action With No Label

low

Blue circular-arrow icon next to transaction rows has no tooltip or legend. New agents must guess or ask what it does. Recognition over recall violation.

Practitioner-Informed Scenarios// 02 — Field research

Practitioner scenarios

Three practitioner-informed scenarios modeling what live observation would surface, a hypothesis set for validation with real ops teams.

Ana — Payments / Cage Agent

Processing pending withdrawal requests at shift start

“I process withdrawals one by one, and every time I hit 'Cancel' I have to pause and think: am I cancelling the withdrawal or cancelling what I'm doing right now?”

Marko — Fraud & Risk Agent

Investigating a declined deposit flagged for review

“Every investigation starts with me un-mangling this JSON before I can even start the actual investigation.”

Ines — AML & Compliance Officer

Verifying a wire transfer account before large release

“Verified just means someone clicked the button. It doesn't mean the data was actually checked.”

Redesign Direction

A pattern already proving its value

The Lifecycle timeline already demonstrates the target, a structured, scannable, timestamped event history replacing opaque JSON. Agents reported faster scanning in early feedback. The recommendation: finish rolling it out, while fixing the critical verification gap and high-severity action-label issues first.

01 / FIX NOW

Cap partial amounts at approved. Validate bank fields before Verify. Clarify withdrawal action labels (Cancel/Decline) with explicit outcome descriptions.

02 / FIX NEXT

Extend Lifecycle pattern to transaction remarks. Introduce bulk withdrawal actions to eliminate one-by-one processing bottleneck.

03 / VALIDATE FIRST

Collect user feedback on redesigned flows with help of rapid prototyping. Validate clarity of new action labels and payment lifecycle with ops agents.

→ Key insight: the highest-leverage change is already prototyped. The Lifecycle pattern (rated Positive in audit) resolves the raw-JSON problem and most scanning-speed issues in one move.

Payments · Redesigned Screens