SafeYelli
Empowering safety through geospatial intelligence and community-driven insights.
Visit safeyelli.inProject at a Glance
SafeYelli is an information website that empowers women to navigate urban India safely through real-time incident reporting, interactive heat mapping, and community-verified safety intelligence.
Every Journey Should Feel Safe
In urban India, millions of women navigate a hidden map of fear every day. Streets that look safe by day transform at night. Popular routes become anxiety-filled journeys. This is the story of how we set out to change that.
The Challenge
No comprehensive geospatial safety data. No real-time incident reporting. No way to know which streets are truly safe.
Our Users
Women aged 18–45 — students, working professionals navigating urban areas — each with unique safety concerns and daily routines.
Our Vision
A geospatial safety platform with real-time reporting, heat mapping, safe route suggestions, and community-verified ratings.
My Contributions to This Project
- Led user research and designed the information architecture
- Created user personas and journey maps from qualitative data
- Designed and built the responsive website interface
- Conducted usability testing and led validation studies across 3 rounds
The Three Personas
Through 45 interviews and 1,247 survey responses, three distinct user archetypes emerged — each with a unique relationship to urban safety.
Common Behavioral Patterns
Despite diverse backgrounds, all three personas share critical safety behaviors that informed every design decision. These aren't edge cases — they're the norm.
Design Implication
- Safety must be the primary lens for every feature — not a secondary filter
- Community and peer trust are more powerful than official data sources
- Time-aware features are essential — safety perception is deeply temporal
- Anonymous contribution lowers barrier to sharing sensitive experiences
Listening to Real Stories
We traveled across Bangalore, talking to women about their daily journeys. What we discovered went far beyond what any dataset could reveal. Every conversation unveiled patterns of fear, resilience, and adaptation.
UX Research Methods
We selected methods across all four quadrants to triangulate findings — combining behavioral observation with attitudinal data, and qualitative depth with quantitative scale.
Key Research Insights
Three findings that shaped everything that followed.
The Numbers Behind the Fear
Our survey of 1,247 women across 12 major metros revealed the scale and shape of the problem — and the appetite for a solution.
Participant Demographics
In Bangalore, 73% of respondents were working professionals.
Feature Priority Rankings
Survey respondents ranked what mattered most — safety intersected every answer.
Significant gap between desired features and current availability highlights untapped market opportunity. Emergency SOS has highest availability (45%) but still falls 51 percentage points short of demand. Real-time safety mapping and route scoring show 80%+ gaps — critical areas for innovation.
Survey Methodology & Validation
How we reached 1,247 respondents and ensured data quality.
Distribution Channels
Quality Assurance
Patterns of Fear & Resilience
Forty-five interviews, conducted in English, Hindi, and Kannada, revealed something no survey could: women have built an entire invisible infrastructure of safety — informal, fragile, and exhausting.
We share safety info in WhatsApp groups. Imagine tapping into thousands of women's experiences.
Community Sharing
Active information sharing via WhatsApp groups and informal networks — but no centralized platform.
Technology Trust
Privacy concerns about location tracking and potential data misuse were the primary barrier.
Time-Dependent Perceptions
Every single participant said safety perception changes dramatically based on time of day.
Route Adaptation
All participants had developed personal mental maps of safe vs. unsafe zones — none of this data was captured anywhere.
Critical Pain Points
Design Opportunities
Interview Methodology
5-step qualitative analysis framework — conducted with 2 independent coders (Cohen's κ = 0.82) to ensure reliability.
Recording & Transcription
All 45 interviews recorded with consent and professionally transcribed verbatim.
Thematic Analysis
Affinity mapping with 2 independent coders — Cohen's κ = 0.82 reliability score.
Quantitative Coding
Themes coded and scored for frequency and intensity across participant groups.
Pattern Recognition
Cross-referencing attitudinal and behavioral data to identify design opportunities.
Validation
Findings validated back with a subset of participants in follow-up sessions.
Interview Guide
Six core question types, each probed 2–3 levels deeper. Interviews conducted in participants' preferred language.
Experience Mapping
"Walk me through a recent journey where you felt unsafe — what did you do differently?"
Tool Usage
"What apps or tools, if any, do you currently use for safety? What's missing?"
Social Behavior
"How do you currently share safety information with friends or family?"
Ideal Scenario
"What features would make you feel more confident navigating your city?"
Privacy Concerns
"What concerns do you have about sharing location or safety data?"
Temporal Context
"How does time of day affect your perception of safety in different areas?"
Designing the Structure
Before a single pixel of UI, we needed to understand how women mentally organize safety information. We used four methods over ten weeks to get this right.
Card Sorting
32 participants organized 65 features through open and closed card sorting to reveal natural categorization patterns.
Mental Models
Mapped cognitive pathways users take when seeking safety information and making navigation decisions.
Tree Testing
150 users completed 8 task scenarios to validate navigation structure and information findability.
Primary Navigation Structure
A flat, 5-tab bottom navigation ensures all critical features are one tap away — no deep menus, no buried safety tools.
User Flow Timing
We designed two critical flows to be fast enough that they never become a friction point in a stressful moment.
IA Testing Results
150 participants completed 8 task scenarios across 3 testing rounds — we iterated until the numbers told us it was right.
Flat navigation outperformed nested menus
Bottom tab bar reduced task completion time by 34% compared to hamburger menu structure.
Emergency features highly discoverable
Users could locate and activate SOS features within 4 seconds across all testing rounds.
Community tab initially confusing
Initial 'Social' label caused confusion — renamed to 'Community' in round 2, success jumped to 89%.
Route comparison highly intuitive
Side-by-side safety score comparison met user expectations without any instruction.
Testing Methodology
Tree Testing
Validated navigation structure without visual design influence to ensure logical hierarchy.
First-Click Testing
Measured user confidence in navigation by analyzing where they clicked first for key tasks.
Task Analysis
Tracked completion rates, time-on-task, and error rates for safety-critical user journeys.
Design Principles from Research
Every design decision traces back to something a participant told us or showed us. These principles became our north star through every iteration.
Information Hierarchy
3 secondsUsers needed immediate context about safety data. We prioritized map visualization above the fold with clear filtering controls — safety information in 3 seconds or less from app open.
Contextual Information
24h time filterUsers consistently asked for time-based context. We built a 24-hour time filter into every safety view, and distance-based filtering to show what's relevant to immediate surroundings.
Trust Through Transparency
Traffic-light systemUsers distrusted black-box safety scores. We introduced a traffic-light system with source attribution and community verification counts — zero learning curve, maximum credibility.
Privacy as a Feature
Anonymous reporting92% would share data if privacy was protected. We made anonymity the default — time, location, and credibility for every data point, with granular privacy controls always visible.
Design System Foundations
Color Palette
Yellow signals trust and community. Red = alert. Green = verified safe.
Typography
Spacing
8px base grid — 8, 16, 24, 32px increments throughout.
Accessibility
All color combinations meet WCAG AA contrast ratios. Icon + label always paired. Critical actions duplicated as text for screen readers.
Homepage
Interactive landing page documenting street harassment patterns in Bengaluru with full bilingual support.
Reporting Page
Anonymous incident reporting system designed to empower community contributions while protecting user privacy.
Community Mapping Initiatives
Ground-up community engagement projects that brought local neighborhoods together to map safety data collaboratively.
Research Outcomes & Impact
The research-driven approach resulted in a product that resonates with user needs and demonstrates measurable impact on women's safety and confidence when navigating urban spaces.
Success Metrics
Quantifiable results demonstrating the platform's impact on user safety and engagement
Key Research Learnings
Critical insights that shaped product decisions and feature development
The geospatial analysis revealed that safety perceptions vary significantly by time of day and day of week. Areas considered safe during daytime often had different safety scores in evenings. This temporal dimension became a critical feature in our heat mapping algorithm.
- • Safety scores fluctuate by up to 40% between day and night
- • Weekends show different patterns than weekdays
- • Transit hubs correlate with higher incident reporting
Diary studies and contextual inquiry uncovered unconscious safety behaviors that users had developed over time. Understanding these existing coping mechanisms helped us design features that augmented rather than replaced learned behaviors.
- • 71% of women choose longer routes if they feel safer
- • Users rely heavily on visual cues (lighting, crowd presence)
- • Sharing location with trusted contacts reduces anxiety by 65%
This project reinforced a belief I now carry into every design engagement: the best interfaces don't feel designed at all. They feel like someone finally understood you. SafeYelli worked because we listened first — and designed second.