Wedge - AI Receipt Itemizer

Project Overview

Managing shared payments, like groceries or restaurant bills, can be awkward, confusing, and prone to error. One person usually pays the full amount, leaving the challenge of dividing costs quickly and accurately.

PROBLEM STATEMENT

Individuals who make purchases for others need a tool to track spending.

SOLUTION

Use AI vision to streamline receipt scanning and easily share spending.

ROLES

Product Designer
UX Researcher & Analyst

RESPONSIBILITIES

Design system, user flow, wireframes, research, branding, and marketing

FREELANCE DURATION

September 2024 - Present

Process

To incorporate brand into a curated collection of reusable components, patterns, and guidelines that ensure a product’s interface stays consistent, scalable, and efficient.
Mapping steps and decisions that show how a user moves through a product’s screens or features to complete specific tasks,
Static representations of a design that showcase detailed visual elements—such as colors, typography, and spacing—before development.
Observing real users interact with the product in context to uncover qualitative data, such as pain points, natural behaviors, and unmet needs.
Afterwards, triaging findings with quantitative data through benchmarking KPIs from surveys, Likert scales, and analysis.
Cycling through feedback and data points to direct design direction in meeting business and user goals.
Serving as a hand-off to engineers to demonstrate and annotate how a design is put together.

Design System

Montserrat, a modern font, hierarchically contrasts well with Inconsolata, a fixed monospace font like receipt text.

Purple evokes a sense of reliability (similar to Zelle)

Yellow complements purple in color theory and aligns
with the brand “lemon wedge” but also money (like 💰)

User Flow

User Research

METHOD

1-1 Observational Usability Test, Post-survey

RESEARCH GOALS

RO1: Obtain early feedback on core user flow

RO2: Assess monetization and marketing strategies

PARTICIPANTS

5 new users with financial means, aged 18–30.

Coded Themes and I-Statements

RO1 - Design Iteration from Qualitative Data

Finding 1. Users wanted to itemize their receipt before entering metadata (if any)

CONSTRAINT

Maintain the transaction requirement of at least 2 members

RESULT

Users start with receipt itemization instead of entering metadata
→ Better aligns with proven, natural human use
→ Clearer button functionality
- “Upload” use case no longer hidden in “Scan Receipt”
- “Entry” renamed to “Manually Enter Receipt”

Finding 2: Users were confused about assigning items to spenders

RESULT

Revisited item assignment flows
→ Users can now select members in 1 tap rather than 2 different taps
→ Global design system change to emphasize interactability via purple accents
→ Finishing the user flow is clearer as a primary CTA, alongside a secondary option to save

Bolt AI - Design Spotlight

I had designer’s block with the new button overlay UI

Generation 1: I had Bolt generate a basic starting point for the features I needed (1) Scan or upload receipt, and (2) Manually enter receipt.
Generation 2: I provided a reference image with more playful circular buttons, attempting to separate the scan and upload buttons.
Generation 3: I ultimately designed something in Figma and fed the PNG image into the prompt to see if it would create something better. The layout was messed up, but I ended up using their color gradient.

RO2 - Marketing from Quantitative Data

FINDINGS

97.5% prefer watching ads to avoid a subscription paywall
- Leading me to suggest non-optional ad placement that doesn’t interfere with subscription conversion rates
84% would recommend the app
- Leading me to suggest invite-based rewards to boost adoption rates through organic lead generation
On the downside, 38% of users would not pay for this app
- Leading me to suggest an empathy-based approach centered around “supporting the team” OR starting our price on the lower end, around our competitor app marketed at $3/month

END RESULT

At our early stage, business goals were to prioritize early adoption rates over profit by tailoring to all types of users:
- Type 1: Free users, limited to 3 scans per month
- Type 2: Subscription users, who want unlimited uses, unlock more automation features, or support the app.
We are valuing our app at $3 per month to be competitive and work towards a sustainable ROI, all while continuing to support free users who value being mindful of their money.