Betterview_proj2_index.png
 

Empowering insurers to take action with Computer Vision Insights.

The final piece of the roof score evolution involves actionable insights. Once we were able to leverage computer vision to diagnose the roof condition (0-100 score), the next step was to figure out how insurers can take action on the issues we flagged. Being able to point to specific areas of the roof and suggest 3 options for taking action was the MVP solution. We provide the insurer with the ability to (1) recommend an action, (2) set rules around continued monitoring of the issue over time, or (3) make the call that no action is required. Ultimately the goal of the insurer is to manage risk and provide the insured with some specific actions they should take to improve the condition of their roof and mitigate risks. Betterview’s role is to flag these issues for insurers, help them decide on a recommendation, and generate a document that can be distributed to the insured to illustrate and explain the action that’s require of them.

 
 

Context

At the core of Betterview is the aerial imagery that’s analyzed, the roof score, and the heat map that illustrates where the issues lie. Our primary goal with the actions system is to help insurance companies leverage our output to make faster and more accurate underwriting decisions. Our secondary goal is to help the insurer communicate what actions the insured needs to take to maintain compliance under their policy.

 
 

Project Goal

How might we make the 0-100 roof score actionable?

My role

Lead research, UX, UI, feedback, and project management

Measures of Success

  • Drive action on policies

  • ROI via risk management and improvement

Process

I worked closely with customers to understand what types of decisions they want to make using Betterview insights. I designed an MVP version of this feature in Spring 2019 and pitched it to a handful of customers to get their feedback. The idea resonated; however, at that point our tech wasn’t able to identify where on the roof the issue was located, which made it difficult to prioritize which issues the insurer should focus on. Nearly a year later I returned to the project because our computer vision advancements can now support a powerful MVP version of this feature. We’re now able to identify which issues exist on the roof and point exactly to the impacted areas as well as diagnose severity via square footage.

This project required working in a cross-functional team over the course of 3 weeks including leadership, design, data science, and engineering. I employed a user-centric, iterative approach to design, develop, and iterate on this crucial piece of the product.

Results

  • Drive action on policies

    • Leverage score to determine internal workflow and which policies require analysis.

    • Use visualization of the score to generate recommended actions and send to the insured. Can then subscribe to the property and monitor compliance.

  • ROI via risk management and improvement

    • We’ve seen 5x ROI by flagging high-risk properties that need (1) a policy adjustment to avoid a likely future loss, and (2) flagging issues that can be resolved to decrease the likelihood of a future loss.