Overview

Transforming bill splitting into a collaborative and seamless group payment experience that removes organizing friction.

Role

Design System Lead

UX/UI Designer

User Researcher

Time

6 weeks

| CONTEXT

The Problem

SnapSplit is a bill-splitting app that lets groups scan receipts and divide costs. I joined as the design lead when the team noticed users were dropping off before completing their bills.

The pattern was consistent. Someone would scan a receipt, items populated automatically, and it felt easy. Then came the assignment stage. The host had to manually assign each item to each person while friends waited. Most users gave up here.

I needed to understand why.

| RESEARCH & DISCOVERY

Understanding the Drop-Off

The team knew users were dropping off, but not why. I ran usability tests with 4 users to find out.

The pattern showed up in every session. Users would scan their receipt, see all the items populate, and feel good about it. Then they'd hit the assignment screen and stop. They'd look at the list, try to figure out who ordered what, attempt to assign a few items, then give up. The interface asked them to go through each item individually and assign it to each person. For a group dinner with shared appetizers and multiple entrees, this turned into dozens of tiny decisions. It was exhausting. None of the 4 users finished.

Watching them struggle, I noticed something else. While the host was stuck in the app trying to figure out assignments, their friends were just... waiting. Doing nothing. The host would look up occasionally, confused and frustrated, trying to remember who ordered what. The collaborative moment of splitting a bill had turned into a solo task.

That's what needed to change.

| USER INSIGHTS

Target Users and Behaviors

Before jumping to solutions, I needed to understand who was actually struggling with this. Our users weren't accountants or business teams. They were young adults splitting dinner with friends, roommates dividing utilities, travelers sorting out trip costs. These were social moments that had turned into admin tasks.

Talking to users revealed what mattered. They wanted to split bills in seconds, not minutes. They wanted the work to feel fair - nobody stuck playing accountant while everyone else scrolled their phone. And they needed to trust the math, especially since SnapSplit shows the split but doesn't handle actual payments.

Speed, fairness, and trust. Those three things became the foundation for everything that followed.

| DESIGN VISION

Making It Collaborative

The solution was clear: stop making one person do all the work. Instead of the host assigning items to everyone, let people claim their own items. Turn bill splitting into something the group does together.

This meant rethinking the entire flow. The host would scan and set up the bill, then share it with the group. Everyone could jump in at the same time, select what they ordered, and see the total update in real time. No more solo burden. No more awkward follow-ups about who owes what.

| THE SOLUTION

Collaborative Assignment

The fix was straightforward: let people claim their own items instead of making the host assign everything. This shifted the burden from one person to the entire group, turning bill splitting into something everyone does together.

| USER FLOW REDESIGN

Solving abandonment at the item assignment step.

The new flow distributes the work across the group while keeping the process simple and visual.

  1. Host Capture

The host scans the receipt. AI extracts all items and costs automatically.

Receipt scanning with automatic item extraction

  1. Verification & Setup

The host quickly verifies items and adds everyone to the bill.

The host verifies the cost data and invites all participants, setting the stage for the collaborative assignment.

  1. Collaborative Assignment (The Fix)

Everyone receives a link and claims their own items simultaneously. No one person is stuck doing all the work.

Each person selects what they ordered

  1. Settlement & Tracking

Final amounts calculate automatically. Everyone sees who's paid and who hasn't.

Payment tracking and final totals

With the redesigned flow complete, we moved into validation to test whether collaborative assignment would actually solve the abandonment problem.

| VALIDATION

Testing the Redesign

We tested high-fidelity prototypes with 6 participants, comparing the legacy flow against the redesigned collaborative experience. While the sample was small, the results were consistent.

| RESULTS

The Results

The legacy flow saw only 2 out of 6 participants complete the bill split. Most got stuck at the assignment step and abandoned the task.

The redesigned collaborative flow changed that. 5 out of 6 participants completed the entire process, a 60% increase in task completion.

What Users Said

The data told one story. User feedback reinforced it:

Users stated that the design helped with the splitting work, making the process feel fair and approachable.

The difference was clear: when the work was shared across the group instead of falling on one person, users stayed engaged. Collaborative assignment eliminated the bottleneck.

| DESIGN SHOWCASE

Design : Before & After

The legacy flow forced the host to manually assign every item. The redesign distributes that work across the entire group.

The Complete Experience

From scan to settlement in four collaborative steps.

The end-to-end experience: scan, verify, collaborate, settle.

|TAKEAWAY

What I Learned

Working as the only designer on a small team taught me to move fast and trust my research. When six users showed the same behavior, I had enough data to make a decision. I didn't need perfect consensus or more testing rounds.

The hardest part was advocating for a solution that required rebuilding the core flow instead of polishing what existed. In a startup, time is everything. Convincing the team to invest in a fundamental change meant showing clear evidence that incremental fixes wouldn't work.

This project taught me that sometimes the best design work isn't visible in the UI. It's in knowing when to challenge the underlying model and having the conviction to see it through.