Overview
I joined Perch as the sole designer to help pivot the company from wildfire sensors to a wildfire risk analysis platform. I led end-to-end product design, translating complex environmental and spatial data into user-friendly tools for communities and institutions.
Working closely with the co-founders and ML team, I helped shape the MVP and build early traction with fire agencies and insurance partners.
Goals:
Turn complex risk data into intuitive, actionable insights.
Design a scalable platform for individuals, agencies, and insurers.
Build a system that encourages risk reduction behaviors.
Founding Product Designer
UX Research, Dashboard Design, Prototyping, Information Architecture, Content Strategy, User Research and Outreach |
CEO, CTO, CFO, Data Scientists, Engineers, Web Developers, Advisors
Q3 2024 – present
Wildfires are becoming more severe. Insurance policies are being cancelled.
Most wildfire mitigation advice is vague or overwhelming. Property owners are being dropped from their insurance plans and don’t know where to start, or how to measure whether they’re improving their risk. Meanwhile, insurance and fire agencies lack scalable tools to communicate mitigation needs.
Helping people act on wildfire risk, not just observe it.
Perch is building a wildfire risk analysis platform: a SaaS product that helps insurers, homeowners, and fire departments assess property-specific risk using a layered scoring system.
To reduce user overwhelm, I organized risk data into three digestible pillars, Predict, Prevent, and Protect, each designed with tailored visualizations and call-to-action components.
Predict
We predict wildfire by using historical data to assess wildfire risks and ignition events
Prevent
We prevent losses by reducing wildfire risk at both household and community levels.
Protect
We protect communities from wildfire by connecting residents to resources to support wildfire preparedness.
The process
Pivoting from sensors to software
Perch began as a hardware startup developing sensors to monitor wildfire risk in hard-to-reach places. By the time I rejoined the team, they had realized that market was saturated and decided to pursue a risk score solution.
I led the end-to-end design process during this pivotal product shift. Throughout this process, I:
Reviewed interview recordings
Synthesized themes
Shared insights with the team
Helped shape the product strategy.
Designing without a finished product
While our data science team worked to develop the algorithm that would drive the platform, I was designing the interface that would present its results. That meant building layouts, hierarchies, and interaction patterns without knowing exactly what the final data would look like.
To manage this uncertainty, I focused on:
Flexibility in structure: designing layouts that could scale with evolving outputs
Tight collaboration with the algorithm team: checking in frequently to align on direction and assumptions
Rooting every design decision in user needs: ensuring clarity and actionability for both technical and non-technical users
Resetting to move forward
As designs progressed, we realized we were missing clarity on who we were building for, and what they actually needed. I paused feature work to realign through user research.
I launched a branching survey and began targeted interviews with homeowners, insurance reps, and fire professionals. Their insights are now shaping the next phase of design.
Lessons from designing in motion
Took ownership of platform design decisions without waiting for full specs.
Collaborated with engineering team to define how data should be visualized.
Prioritized feedback from non-technical users to simplify onboarding.
What’s ahead for Perch
We’re now preparing to test our prototype with homeowners, fire professionals, and insurance reps. These sessions will help us validate how well users understand their wildfire risk score and whether the design prompts actionable next steps.
Based on this feedback, I’ll refine the interaction model, simplify the onboarding flow, and develop our high-fidelity designs for pitching and pilot deployment.