Overview
Problem
Goal
Solution
Research
Users & Context
Discovery & Definition
Design & Iteration
Delivery
Reflection

Empowering Shoppers to Identify Harmful Ingredients
through Transparent Data Scanning

Bye, Bad Tox.

Health & Wellness

Product Design

User Research

5-6 min read

My role:

Product Designer(IA, Interaction, Visuals, Content Strategy)

My role:

Product Designer

Team:

Solo Project

Tools:

Figma, Maze, Balsamiq, Lovable

Duration:

6 Months

Overview

Introduction

Bye, Bad Tox. is a mobile tool that helps shoppers make decisions quickly and confidently when purchasing personal care products. The mission is simple: turn complex labels into plain-language guidance through ingredient transparency.

Introduction

Bye, Bad Tox. is a mobile tool that helps shoppers make decisions quickly and confidently when purchasing personal care products. The mission is simple: turn complex labels into plain-language guidance through ingredient transparency.

Introduction

Bye, Bad Tox. is a mobile tool that helps shoppers make decisions quickly and confidently when purchasing personal care products. The mission is simple: turn complex labels into plain-language guidance through ingredient transparency.

My Role

I was the solo product designer, leading research, IA, visual design, UX writing, and prototyping—owning the end-to-end experience and user testing.

My Role

I was the solo product designer, leading research, IA, visual design, UX writing, and prototyping—owning the end-to-end experience and user testing.

My Role

I was the solo product designer, leading research, IA, visual design, UX writing, and prototyping—owning the end-to-end experience and user testing.

Problem

Problem #1

Label Literacy

Shoppers can’t decode INCI terms, tiny text, or marketing claims; “natural” is often equated with “safe.”

“

I prefer simple ingredients that I can understand—not chemical names like ‘acrylic acid.’

“

I prefer simple ingredients that I can understand—not chemical names like ‘acrylic acid.’

“

I prefer simple ingredients that I can understand—not chemical names like ‘acrylic acid.’

Problem #2

Fragmented Information

Trustworthy safety information is scattered across unverified platforms, causing decision delay and consumer hesitation.

“

A lot of info comes from Google and I’m not sure if it’s accurate. It’d be better if a trusted public agency compiled ingredient and hazard info.

“

A lot of info comes from Google and I’m not sure if it’s accurate. It’d be better if a trusted public agency compiled ingredient and hazard info.

“

A lot of info comes from Google and I’m not sure if it’s accurate. It’d be better if a trusted public agency compiled ingredient and hazard info.

Defining the problem

“

How might we deliver plain, trustworthy, and personalized guidance at the moment of choice—so users can make quick, confident, and safe purchases without deep research?

“

How might we deliver plain, trustworthy, and personalized guidance at the moment of choice—so users can make quick, confident, and safe purchases without deep research?

“

How might we deliver plain, trustworthy, and personalized guidance at the moment of choice—so users can make quick, confident, and safe purchases without deep research?

Problem Severity

64.3%

of users have expressed profound concerns regarding the health impacts of specific ingredients

Problem Severity

64.3%

of users have expressed profound concerns regarding the health impacts of specific ingredients

Problem Severity

64.3%

of users have expressed profound concerns regarding the health impacts of specific ingredients

Action Willingness

81.0%

of users have demonstrated a strong inclination to modify their purchasing behaviors and habits

Action Willingness

81.0%

of users have demonstrated a strong inclination to modify their purchasing behaviors and habits

Action Willingness

81.0%

of users have demonstrated a strong inclination to modify their purchasing behaviors and habits

Goal

EXPERIENCE GOALS

Deliver at the moment value

Center the app on scan/search to present a clear ingredient summary, risk rating and product review within one screen, enabling quick, confident choices.

EXPERIENCE GOALS

Deliver at the moment value

Center the app on scan/search to present a clear ingredient summary, risk rating and product review within one screen, enabling quick, confident choices.

EXPERIENCE GOALS

Deliver at the moment value

Center the app on scan/search to present a clear ingredient summary, risk rating and product review within one screen, enabling quick, confident choices.

Scalable, consistent system

Build a reusable component library and a shared content taxonomy to ensure consistency, allowing faster iteration with less work.

Scalable, consistent system

Build a reusable component library and a shared content taxonomy to ensure consistency, allowing faster iteration with less work.

Scalable, consistent system

Build a reusable component library and a shared content taxonomy to ensure consistency, allowing faster iteration with less work.

EXPERIENCE GOALS

Increase content relevancy

Use personalization to display the most relevant guidance and products—so users don’t need to compare or research themselves.

EXPERIENCE GOALS

Increase content relevancy

Use personalization to display the most relevant guidance and products—so users don’t need to compare or research themselves.

EXPERIENCE GOALS

Increase content relevancy

Use personalization to display the most relevant guidance and products—so users don’t need to compare or research themselves.

Unify fragmented information

Users often switch between different tools and platforms to evaluate products. This app aims to bring education, safety insights, peer reviews, professional advice, and product discovery all in one place.

Unify fragmented information

Users often switch between different tools and platforms to evaluate products. This app aims to bring education, safety insights, peer reviews, professional advice, and product discovery all in one place.

Unify fragmented information

Users often switch between different tools and platforms to evaluate products. This app aims to bring education, safety insights, peer reviews, professional advice, and product discovery all in one place.

Solution

Bye Bad Tox identifies harmful toxins in personal care labels, translates complex chemical terms into plain-language safety ratings, and provides personalized product recommendations.

By leveraging machine learning to build a unique skin profile, the app tracks your sensitivity patterns to ensure every purchase aligns with your long-term wellness goals.

Key features

Scan or Snap

Scan a barcode or snap a photo to upload and identify products instantly.Can’t find it? Add it manually, submit for review to help grow the database, and earn coins along the way.

Scan or Snap

Scan a barcode or snap a photo to upload and identify products instantly.Can’t find it? Add it manually, submit for review to help grow the database, and earn coins along the way.

Scan or Snap

Scan a barcode or snap a photo to upload and identify products instantly.Can’t find it? Add it manually, submit for review to help grow the database, and earn coins along the way.

Product Detail, Explained

Each product receives a risk rating (toxicity, irritancy, allergy): Low / Medium / High, based on published research.

Product Detail, Explained

Each product receives a risk rating (toxicity, irritancy, allergy): Low / Medium / High, based on published research.

Product Detail, Explained

Each product receives a risk rating (toxicity, irritancy, allergy): Low / Medium / High, based on published research.

Ingredient Encyclopedia

A deep dive for every ingredient: an effectiveness radar, usage tips, and cautions.

Explore related articles and buying advice to help you decide with clarity.

Ingredient Encyclopedia

A deep dive for every ingredient: an effectiveness radar, usage tips, and cautions.

Explore related articles and buying advice to help you decide with clarity.

Ingredient Encyclopedia

A deep dive for every ingredient: an effectiveness radar, usage tips, and cautions.

Explore related articles and buying advice to help you decide with clarity.

Expert Articles & Quizzes

Learn from professionals and trusted-organization articles that connect science to daily choices.

Take a quick quiz at the end to reinforce what you learned.

Expert Articles & Quizzes

Learn from professionals and trusted-organization articles that connect science to daily choices.

Take a quick quiz at the end to reinforce what you learned.

Expert Articles & Quizzes

Learn from professionals and trusted-organization articles that connect science to daily choices.

Take a quick quiz at the end to reinforce what you learned.

Community Reviews

Get real-world experiences—honest reviews, Q&A, and favorite picks.

Join the discussion to share recommendations, tips, and discover what actually works for you.

Community Reviews

Get real-world experiences—honest reviews, Q&A, and favorite picks.

Join the discussion to share recommendations, tips, and discover what actually works for you.

Community Reviews

Get real-world experiences—honest reviews, Q&A, and favorite picks.

Join the discussion to share recommendations, tips, and discover what actually works for you.

Research

Conducted mixed-methods research—including a literature review, competitive analysis, a survey (n=169), and 7 deep-dive interviews—to uncover the gaps in label literacy.

The "Natural" Myth

Consumers equate "natural" with "safe" but lack the literacy to decode INCI terms. Over 71% of surveyed users struggle with complex chemical names, leading to mismatched purchases.

The "Natural" Myth

Consumers equate "natural" with "safe" but lack the literacy to decode INCI terms. Over 71% of surveyed users struggle with complex chemical names, leading to mismatched purchases.

The "Natural" Myth

Consumers equate "natural" with "safe" but lack the literacy to decode INCI terms. Over 71% of surveyed users struggle with complex chemical names, leading to mismatched purchases.

“

The products bought at large stores usually have English labels. They list a lot of ingredients, but you don't know which ones are safe or unnecessary.

“

The products bought at large stores usually have English labels. They list a lot of ingredients, but you don't know which ones are safe or unnecessary.

“

The products bought at large stores usually have English labels. They list a lot of ingredients, but you don't know which ones are safe or unnecessary.

Peer Trust > Official Ads

Social validation is the primary trust signal. Users rely heavily on peer reviews (72%) and expert endorsements (e.g., dermatologists) rather than brand-led marketing.

Peer Trust > Official Ads

Social validation is the primary trust signal. Users rely heavily on peer reviews (72%) and expert endorsements (e.g., dermatologists) rather than brand-led marketing.

Peer Trust > Official Ads

Social validation is the primary trust signal. Users rely heavily on peer reviews (72%) and expert endorsements (e.g., dermatologists) rather than brand-led marketing.

“

I'd prefer brands like La Roche-Posay that cooperate with dermatologists, because I feel like the doctor has given their endorsement.

“

I'd prefer brands like La Roche-Posay that cooperate with dermatologists, because I feel like the doctor has given their endorsement.

“

I'd prefer brands like La Roche-Posay that cooperate with dermatologists, because I feel like the doctor has given their endorsement.

Fragmented & Memory-Based

Safety tracking is currently a manual, high-effort task. Most tools are too text-heavy for quick in-store use, forcing users to rely on memory for allergy triggers.

Fragmented & Memory-Based

Safety tracking is currently a manual, high-effort task. Most tools are too text-heavy for quick in-store use, forcing users to rely on memory for allergy triggers.

Fragmented & Memory-Based

Safety tracking is currently a manual, high-effort task. Most tools are too text-heavy for quick in-store use, forcing users to rely on memory for allergy triggers.

Users & Context

The Ritualist

"Scent is my personal sanctuary to recenter, not a performance."

JOBS-TO-BE-DONE

When studying, journaling, or winding down, choose a scent that sets the mood and helps her feel grounded.

Pain points

Overwhelmed by scientific jargon and anxious about hidden irritants that trigger sensory discomfort.

The Ritualist

"Scent is my personal sanctuary to recenter, not a performance."

JOBS-TO-BE-DONE

When studying, journaling, or winding down, choose a scent that sets the mood and helps her feel grounded.

Pain points

Overwhelmed by scientific jargon and anxious about hidden irritants that trigger sensory discomfort.

The Ritualist

"Scent is my personal sanctuary to recenter, not a performance."

JOBS-TO-BE-DONE

When studying, journaling, or winding down, choose a scent that sets the mood and helps her feel grounded.

Pain points

Overwhelmed by scientific jargon and anxious about hidden irritants that trigger sensory discomfort.

The Pragmatist

"Spraying perfume is the final piece of my outfit—it makes me feel complete."

JOBS-TO-BE-DONE

Before heading out to social events, pick a subtle, reliable scent to feel prepared.

Pain points

Skeptical of abstract marketing claims and fears that low-quality ingredients will cause headaches.

The Pragmatist

"Spraying perfume is the final piece of my outfit—it makes me feel complete."

JOBS-TO-BE-DONE

Before heading out to social events, pick a subtle, reliable scent to feel prepared.

Pain points

Skeptical of abstract marketing claims and fears that low-quality ingredients will cause headaches.

The Pragmatist

"Spraying perfume is the final piece of my outfit—it makes me feel complete."

JOBS-TO-BE-DONE

Before heading out to social events, pick a subtle, reliable scent to feel prepared.

Pain points

Skeptical of abstract marketing claims and fears that low-quality ingredients will cause headaches.

Discovery & Definition

Ideation

Synthesized 107 raw ideas into 8 functional clusters through affinity mapping. To ensure a user-anchored solution, I prioritized features that directly bridge the gap in label literacy:

  • Personalization and smart recommendations

  • Clear, scannable presentation to reduce cognitive load.

  • Credible sources and transparent reasoning to build trust.

  • Community participation and rewarding system to sustain engagement.

Ideation

Synthesized 107 raw ideas into 8 functional clusters through affinity mapping. To ensure a user-anchored solution, I prioritized features that directly bridge the gap in label literacy:

  • Personalization and smart recommendations

  • Clear, scannable presentation to reduce cognitive load.

  • Credible sources and transparent reasoning to build trust.

  • Community participation and rewarding system to sustain engagement.

Ideation

Synthesized 107 raw ideas into 8 functional clusters through affinity mapping. To ensure a user-anchored solution, I prioritized features that directly bridge the gap in label literacy:

  • Personalization and smart recommendations

  • Clear, scannable presentation to reduce cognitive load.

  • Credible sources and transparent reasoning to build trust.

  • Community participation and rewarding system to sustain engagement.

Information Architecture & Tree Testing

I drafted the sitemap covering main flows, then iterated the labels and groupings after testing.

Information Architecture & Tree Testing

I drafted the sitemap covering main flows, then iterated the labels and groupings after testing.

Information Architecture & Tree Testing

I drafted the sitemap covering main flows, then iterated the labels and groupings after testing.

Validation

  1. Qualitative pilots (n=5) — in-person tree tests followed by short interviews to remove big blockers.

  2. Unmoderated tree testing (n=104) — quantified findability, misroutes, and wording comprehension for tasks. 

Validation

  1. Qualitative pilots (n=5) — in-person tree tests followed by short interviews to remove big blockers.

  2. Unmoderated tree testing (n=104) — quantified findability, misroutes, and wording comprehension for tasks. 

Validation

  1. Qualitative pilots (n=5) — in-person tree tests followed by short interviews to remove big blockers.

  2. Unmoderated tree testing (n=104) — quantified findability, misroutes, and wording comprehension for tasks. 

optimization

  1. Revised vague microcopy and similar labels that caused initial misroutes.

  2. Optimized copy to support "skimming" behavior, ensuring key information stands out during task navigation.

  3. Bridged the gap between user expectations and system hierarchy through iterative testing.

For depth vs. density, I implemented a shallower hierarchy to meet local user expectations. Used progressive disclosure and multiple entry points to maintain clarity without burying essential details.

optimization

  1. Revised vague microcopy and similar labels that caused initial misroutes.

  2. Optimized copy to support "skimming" behavior, ensuring key information stands out during task navigation.

  3. Bridged the gap between user expectations and system hierarchy through iterative testing.

For depth vs. density, I implemented a shallower hierarchy to meet local user expectations. Used progressive disclosure and multiple entry points to maintain clarity without burying essential details.

optimization

  1. Revised vague microcopy and similar labels that caused initial misroutes.

  2. Optimized copy to support "skimming" behavior, ensuring key information stands out during task navigation.

  3. Bridged the gap between user expectations and system hierarchy through iterative testing.

For depth vs. density, I implemented a shallower hierarchy to meet local user expectations. Used progressive disclosure and multiple entry points to maintain clarity without burying essential details.

Design & Iteration

Wireframes

I produced wireframes to quickly visualize ideas for testing and feedback. Here are some early screens.

Wireframes

I produced wireframes to quickly visualize ideas for testing and feedback. Here are some early screens.

Wireframes

I produced wireframes to quickly visualize ideas for testing and feedback. Here are some early screens.

Usability Testing

Think Aloud Walkthrough

Conducted two rounds of task-based walkthroughs (Low-fi & Hi-fi) to evaluate product discovery, scanning logic, and the content learning loop.

Findings

Problem

  • Users expected product search/recommendations first on Homepage, not articles.

  • Ambiguous icon/field labels reduced first clicks.

  • Quizzes lacked immediate feedback.

Pivot

  • Prioritized search entry and re-ordered homepage modules.

  • Unified search UI and added example queries to guide users.

  • Implemented instant results and direct links to buying guidance.

Think Aloud Walkthrough

Conducted two rounds of task-based walkthroughs (Low-fi & Hi-fi) to evaluate product discovery, scanning logic, and the content learning loop.

Findings

Problem

  • Users expected product search/recommendations first on Homepage, not articles.

  • Ambiguous icon/field labels reduced first clicks.

  • Quizzes lacked immediate feedback.

Pivot

  • Prioritized search entry and re-ordered homepage modules.

  • Unified search UI and added example queries to guide users.

  • Implemented instant results and direct links to buying guidance.

Think Aloud Walkthrough

Conducted two rounds of task-based walkthroughs (Low-fi & Hi-fi) to evaluate product discovery, scanning logic, and the content learning loop.

Findings

Problem

  • Users expected product search/recommendations first on Homepage, not articles.

  • Ambiguous icon/field labels reduced first clicks.

  • Quizzes lacked immediate feedback.

Pivot

  • Prioritized search entry and re-ordered homepage modules.

  • Unified search UI and added example queries to guide users.

  • Implemented instant results and direct links to buying guidance.

Iteration

Decision 1: Dropping the AI Helper

Insight

During testing, users couldn't locate the AI assistant, and when they did, they questioned whether its recommendations were accurate. At the moment of purchase, they needed immediate answers, not a conversation.

Decision

Replaced the chat interface with algorithmic recommendations and a clear preference setup.

Impact

Reduced user detour and created a faster path to relevant products.

Decision 1: Dropping the AI Helper

Insight

During testing, users couldn't locate the AI assistant, and when they did, they questioned whether its recommendations were accurate. At the moment of purchase, they needed immediate answers, not a conversation.

Decision

Replaced the chat interface with algorithmic recommendations and a clear preference setup.

Impact

Reduced user detour and created a faster path to relevant products.

Decision 1: Dropping the AI Helper

Insight

During testing, users couldn't locate the AI assistant, and when they did, they questioned whether its recommendations were accurate. At the moment of purchase, they needed immediate answers, not a conversation.

Decision

Replaced the chat interface with algorithmic recommendations and a clear preference setup.

Impact

Reduced user detour and created a faster path to relevant products.

Decision 2: Prioritizing "Scan & Search"

Insight

Users couldn't find the scan feature because its placement conflicted with their existing mental models. The app's most important action wasn't salient enough.

Decision

Unified the Search UI at the top-right and added visual hints.

Impact

More participants successfully reached product details on the first click.

Decision 2: Prioritizing "Scan & Search"

Insight

Users couldn't find the scan feature because its placement conflicted with their existing mental models. The app's most important action wasn't salient enough.

Decision

Unified the Search UI at the top-right and added visual hints.

Impact

More participants successfully reached product details on the first click.

Decision 2: Prioritizing "Scan & Search"

Insight

Users couldn't find the scan feature because its placement conflicted with their existing mental models. The app's most important action wasn't salient enough.

Decision

Unified the Search UI at the top-right and added visual hints.

Impact

More participants successfully reached product details on the first click.

Decision 3: Contextualizing Risk

Insight

During testing, no one found Safety explanations buried in Settings. Users instinctively looked for risk information on the product detail page, where the decision was actually happening.

Decision

Moved risk rationales inline on the product detail page using a color-coded hierarchy.

Impact

Higher engagement with safety content during the decision-making process.

Decision 3: Contextualizing Risk

Insight

During testing, no one found Safety explanations buried in Settings. Users instinctively looked for risk information on the product detail page, where the decision was actually happening.

Decision

Moved risk rationales inline on the product detail page using a color-coded hierarchy.

Impact

Higher engagement with safety content during the decision-making process.

Decision 3: Contextualizing Risk

Insight

During testing, no one found Safety explanations buried in Settings. Users instinctively looked for risk information on the product detail page, where the decision was actually happening.

Decision

Moved risk rationales inline on the product detail page using a color-coded hierarchy.

Impact

Higher engagement with safety content during the decision-making process.

Delivery

Building the Brand

I created a design system for consistency and accessibility. The overall design maintains a clean, minimal aesthetic that reinforces the brand identity, to convey freshness, safety, and ease.

Building the Brand

I created a design system for consistency and accessibility. The overall design maintains a clean, minimal aesthetic that reinforces the brand identity, to convey freshness, safety, and ease.

Building the Brand

I created a design system for consistency and accessibility. The overall design maintains a clean, minimal aesthetic that reinforces the brand identity, to convey freshness, safety, and ease.

Validation Metrics

  • Usability score: 85/100

  • Misreading Rate Reduction: ~40% (fewer incorrect picks caused by label confusion)

Validation Metrics

  • Usability score: 85/100

  • Misreading Rate Reduction: ~40% (fewer incorrect picks caused by label confusion)

Validation Metrics

  • Usability score: 85/100

  • Misreading Rate Reduction: ~40% (fewer incorrect picks caused by label confusion)

Final Design

View prototype

Reflection

Learning

Without a clear goal and decision criteria, it was easy to chase feedback or prettier patterns, which led to edits that weren’t grounded in evidence. Changes should tied to research insights, a usability issue, or a specific metric—so scope stays steady and the design stays consistent.

Learning

Without a clear goal and decision criteria, it was easy to chase feedback or prettier patterns, which led to edits that weren’t grounded in evidence. Changes should tied to research insights, a usability issue, or a specific metric—so scope stays steady and the design stays consistent.

Learning

Without a clear goal and decision criteria, it was easy to chase feedback or prettier patterns, which led to edits that weren’t grounded in evidence. Changes should tied to research insights, a usability issue, or a specific metric—so scope stays steady and the design stays consistent.

Constraint

Solo work

Course restrictions had me to work solo. This taught me to own end-to-end product experience, but it means fewer critiques, more blind spots, and no division of labor. Collaboration would have brought fresh perspectives and leveraged individual strengths.

Solo work

Course restrictions had me to work solo. This taught me to own end-to-end product experience, but it means fewer critiques, more blind spots, and no division of labor. Collaboration would have brought fresh perspectives and leveraged individual strengths.

Solo work

Course restrictions had me to work solo. This taught me to own end-to-end product experience, but it means fewer critiques, more blind spots, and no division of labor. Collaboration would have brought fresh perspectives and leveraged individual strengths.

limited metrics

The product wasn’t fully developed and shipped, so I couldn’t track real conversion or retention rate. A working prototype would allow field tests and instrumentation for decision-quality and repeat use.

limited metrics

The product wasn’t fully developed and shipped, so I couldn’t track real conversion or retention rate. A working prototype would allow field tests and instrumentation for decision-quality and repeat use.

limited metrics

The product wasn’t fully developed and shipped, so I couldn’t track real conversion or retention rate. A working prototype would allow field tests and instrumentation for decision-quality and repeat use.

Future

  • Expand sample size and diversity, and run field tests (in-store and during online checkout) to measure first-click rate, time-to-decision, and abandonment.

  • Ship a MVP to wire up the product/ingredient database and basic personalization.

Future

  • Expand sample size and diversity, and run field tests (in-store and during online checkout) to measure first-click rate, time-to-decision, and abandonment.

  • Ship a MVP to wire up the product/ingredient database and basic personalization.

Future

  • Expand sample size and diversity, and run field tests (in-store and during online checkout) to measure first-click rate, time-to-decision, and abandonment.

  • Ship a MVP to wire up the product/ingredient database and basic personalization.

Create a free website with Framer, the website builder loved by startups, designers and agencies.