6 mins read
Enterprise Workflow that Handles $100M+ in Transactions
Reducing operational dependency and transforming a fragmented multi-tool process
into a guided enterprise experience.
Business Impact
Received the "Pat On The Back" award in recognition of the project's operational and business impact.
My Contribution
Ideation
Research
Strategy
Cross functional alignment
High fidelity design
End to end design ownership.
Led 6 out of 12 research sessions
Established design review processes to ensure a smooth hand-off and to stay on track.
Project timeline
3 months
Team
Me, 1 PM, 1 UX researcher, Engineers +more
Entering the World of Enterprise
The Journey Behind Every Transaction
When a company needs thousands of software licenses for its employees,
they don't simply visit a website and click "Buy Now".
Click to enlarge
What happens when something goes wrong after an order
is placed?
There are around 20 reasons why a credit request will be raised
Introducing credits
Credits are used to correct orders after they have been processed. Customer
reaches out to operations when they have to raise a request.
The value can either be:
Refunded to the customer, or
Applied towards another product (Re-billing: To purchase something
else instead of returning the money.
The process of creating & managing these credits is the focus of this case study.
Credit Request Journey
Act 1 — The Business Problem
While credits were critical to financial operations, the UX supporting them
had not evolved with the growing complexity of the business.
Click to enlarge
Act 2 — Discovery & Ground Research
Business understood the outcomes, but not necessarily the root cause. To understand
why requests were slow, costly, and error-prone, I along with a UX researcher conducted
research sessions with the people using the workflow every day.
Click to enlarge
How we used AI: We created a small AI app that analyzed, created themes and
categorized support tickets that helped in figuring out a connection of what credit
reasons lead to what downstream problems and figure out our P0s, P1s and P2s.
Act 3 — Friction Points We Found
🧠 Reliance on Tribal Knowledge:
Critical workflow rules lived only with experienced staff. When they left, teams
had to rely on undocumented processes (High employee turnaround).
🧩 Multiple Disconnected Tools:
Depending on the kind of reasons for the credit request, users had to navigate
several tools to complete a single request, often creating confusion and inefficiencies.
⚠️ Frequent Request Rejections:
Due to lack of tribal knowledge - Missing steps, lack of accuracy and
inconsistent execution led to avoidable rejections & rework
💻 Complex, Error-Prone Workflow:
Limited validation and disconnected systems made mistakes easy to make
and difficult to prevent.
Where Things Became Complex
Over 20 credit reasons existed, so the team aligned on a phased rollout, prioritizing a subset of reasons in each release. The challenge was deciding what belonged in Phase 1, as there was little to no alignment between teams.
Shift in Strategy
Solving individual credit reasons wasn't scalable. I noticed that many requests
shared the same workflow patterns, and operational tasks.I identified an opportunity to create a scalable capability that delivered
value across multiple credit reasons.This shifted the conversation from "Which reason should we fix first?"
to "Building what capabilities can impact most reasons?"
Core UX principles
The research uncovered recurring patterns across the workflow. Rather than solving individual pain
points, I focused on principles that could systematically reduce complexity across the entire workflow.
Key UX decisions
Guided by research and UX principles, these decisions transformed a
fragmented process into a streamlined workflow.
Automation based on credit reason
Problem:
Operations teams manually gathered supporting documents from multiple systems before submitting a credit request. Missing attachments were a common cause of avoidable rejections.
Decision:
Instead of treating document collection as a user task, I explored opportunities
to automate it within the workflow.
How I Got There:
Cross-functional mapping: I facilitated discussions with Product and Engineering
to map how data flowed across Quotes, Contracts, Reports, and Orders.Automated retrieval: This revealed opportunities to automatically retrieve and attach relevant documentation from these workspaces based on the selected credit reason.
Why This Approach:
Scalable capability: A single automation capability could support multiple credit reasons, reducing manual effort while avoiding the need for custom solutions for each scenario.
Metrics:
75% reduction in rejections due to missing documents. The reason this metric is so high is because the solution I proposed shifted the baseline, leaving external, customer-provided files as the only remaining source of document failure.
System level thinking drove very high impact results for this problem.
Click to enlarge
Guided request creation
Problem:
System limitations: While automated uploads worked for basic credit reasons, other scenarios required external supporting documents that didn't exist within the system.
Process dependencies: Beyond missing files, certain requests required users to complete complex manual tasks like executing a rebill prior to submission to guarantee approval.
Downstream risk: Leaving these gaps unguided would lead to missing documentation,
inconsistent submissions, and avoidable rejections.
Decision:
Contextual guidance: I introduced contextual guidance immediately after a credit reason was selected, surfacing the specific documents and steps required for that request.
Why This Approach:
Just-in-time guidance: The goal was to guide users at the point of action
rather than expecting them to remember process requirements.Process standardization: This helped standardize request creation across
both experienced and new team members.
Click to enlarge
Constraint:
Operational flexibility: Legitimate business exceptions existed where
a request could be approved without mandatory documents. Because
of these edge cases, hard-blocking the user from submitting was impossible.
Design Response:
Soft validation fences: Instead of enforcing hard validation, I added a final confirmation step highlighting any missing documents or incomplete actions before submission, allowing users to make an informed decision while preserving operational flexibility.
Click to enlarge
Outcome:
Users received reason-specific guidance throughout the workflow, reducing dependency on tribal knowledge, improving submission quality without disrupting legitimate edge cases.
Metrics:
27% success in Standard Requests. Contextual guidance proved highly effective for straightforward credit request types.
The Phase 2 Roadmap: While we anticipated a higher overall lift based on our automation success, complex credit exceptions proved that static on-screen guidance wasn't enough for highly variable edge cases. We have scoped deeper system optimizations for these specific edge cases in our Phase 2 roadmap.
Dual-Path Invoice Initiation
Problem:
Selecting the correct invoice was critical, yet a common source of errors.
Large customers often had multiple invoices across products and departments, making verification difficult in legacy systems.
Decision:
I introduced two initiation paths: a quick entry point from Orders and a verification-focused entry point from Order Details for more complex scenarios.
Click to enlarge
Why This Approach:
Usability discovery: During prototype testing, I asked users how they would raise a credit from a familiar orders home screen they are used to seeing. While some noticed the new "Initiate credit action" on the home screen (Fig.1), many said they would search for the order, verify it there to ensure they don't make a mistake then raise a credit.
Dual-user needs: For most orders, the key identification information shown in the initiate credit action panel would suffice, but complex orders required detailed information that is only available in the order details page.
Uncovering tribal knowledge: This led to creating another initiation action on that screen (Fig.1.2). This was also one of those cases where tribal knowledge came into play and we didn't know such a problem existed, but discovered it during the first prototype testing.
Outcome:
Credit request rejections due to wrong invoice selection reduced drastically.
Metrics: ~7% reduction in rejections due to wrong invoice selection
Fig.1.2
Visual Cues
Problem:
In complex enterprise workflows, modifying data across large layouts makes it easy for users to lose track of exactly which fields they changed.
Because these systems handle high-value transactions, missing an accidental modification or failing to double-check an input before clicking submit poses a massive financial risk.
Decision:
I introduced real-time visual tracking for edited states: a subtle yellow background fill immediately highlights any modified field, paired with a global "Show edited only" filter toggle to facilitate an intentional, pre-submission review.
Why this approach:
Accuracy over speed: In enterprise platforms handling high-value transactions, preventing a single data-entry error is infinitely more critical than shaving a few seconds off the workflow timeline.
Reduce cognitive fatigue: The persistent yellow highlights act as instant visual signposts, completely removing the need for users to rely on memory to recall what they altered.
Isolating the variables: The "Show edited only" toggle instantly filters out untouched system data, collapsing dense pages into a clean, focused ledger of changes for a flawless final audit.
Outcome:
Users gained complete visibility over their modifications, transforming an anxious, blind submission process into a high-confidence checkout validation.
Click to enlarge
Phase 2 Roadmap
While we were happy about the reduction in credit request rejections and cost saved. One of the questions I had asked when we started this project was - No company is happy with giving refunds and credits. The goal is to have minimal to no credit requests in the first place.
While the team agreed on this, it currently wasn't the P0.I had pitched a few ideas to reduce credit requests from the customer side. Would look to explore that next.
Strategic Takeaways
The best interface is system automation: This project proved that true enterprise UX isn't about making forms prettier; it's about shifting the burden of labor from the human to the machine.
By leading cross-functional data-mapping sessions, we wiped out the core friction point before the user even encountered the screen.
Thanks for reading!