GGenpact2 min read
Redesigning enterprise invoice processing for 100,000+ users.
As solo UX designer, I rebuilt Genpact's Salesforce-based accounts payable platform for Fortune 500 clients — introducing user-centered design to an organisation that had never worked that way before.
- Role
- Solo UX Designer
- Scale
- 100K+ users, 30+ accounts
- Platform
- Salesforce (AP Flow)
- Year
- 2020
Challenge
An accounts-payable platform used across 30+ Fortune 500 accounts made exception handling slow, manual and error-prone.
Strategy
Working solo, I removed context switching, surfaced mismatches automatically, and made the case for design practice inside an engineering-led organisation.
Results
67% turnaround reduction and error rates from 45% to 15%, on a platform serving 100K+ users.
overview
Overview
AP Flow was Genpact's accounts payable platform for Fortune 500 companies, built on Salesforce. Invoice processors were losing time to constant context-switching between systems, manual mismatch checks, and a workflow that hadn't been designed around how the work actually happened — it had grown around what the platform could technically do.
Working solo, without an existing design practice to lean on, I owned research, design, and stakeholder buy-in end to end — and had to make the case for design thinking itself along the way.
research
Research
I spent 35+ hours observing invoice processors directly and interviewed 20+ of them across accounts to understand where time and accuracy were actually being lost — not where the platform's original spec assumed it would be lost.
- Processors were manually cross-referencing purchase orders against invoice line items in separate windows.
- Mismatches were caught late, after significant manual re-checking, rather than flagged as they occurred.
- Multi-PO invoices had no clear signal for which purchase order was primary.
- Context — scroll position, selected line items — was lost constantly during review, forcing re-work.
Two systems, 1,000+ line items to check by hand, and 30–40 minutes for every complex invoice.
Zero real-time sync between the processor and buyer portals — so both sides argued about stale data.
Watching the work, not the wireframe
Thirty-five hours beside invoice processors and twenty-plus interviews showed the time was not being lost where the platform spec assumed. It went on cross-referencing purchase orders against line items in two windows, catching mismatches late, guessing which purchase order was primary on multi-PO invoices, and losing scroll position and selection on every return trip.
solution
Solution
The redesign centered on removing context-switching and surfacing exceptions automatically, rather than relying on processors to catch them manually.
- Side-by-side PO verification — purchase order and invoice shown together, eliminating window-switching.
- Automated mismatch detection — line-item discrepancies flagged with visual indicators instead of manual comparison.
- Real-time sync between the Invoice Processor and Buyer portals, removing stale-state confusion.
- Intelligent primary PO identification for multi-purchase-order invoices.
- Persistent context preservation while scrolling through line items.
- Visual exception highlighting using consistent colour coding for fast triage.
impact
Impact
The redesigned platform contributed to AP Flow becoming one of Genpact's highest-revenue products — and proved the value of design thinking to an organisation that hadn't previously invested in it.
How this was measured
Processing time and error rates are client-reported operational figures from the accounts the platform ran on, compared before and after rollout. The annual saving is a business estimate calculated from those processing figures by the delivery team, not an independently audited finance number. Active users describe the platform's reach across 30+ accounts, which is scale rather than design impact on its own — and revenue performance also reflected commercial, delivery and account decisions beyond design.
What I would qualify
The processing-time and error-rate figures are client-reported operational numbers, compared before and after rollout. The annual saving is a business estimate derived from them by the delivery team, not an audited finance figure — and the platform's commercial performance reflected account, delivery and pricing decisions well beyond design.