Restimate

Timeline: 1 year, January 2021 - March 2022

Role: UI Design, User Researcher, Content Designer, Developer

Tools: Miro, Retool, HTML/CSS, SQL

Restimate is a NSF-funded research project that aims to develop new methods to estimate disaster recovery times of critical infrastructures, such as electricity and water systems, after a natural disaster.

CASE STUDY

There is no standard practice for estimating disaster recovery times currently.

Resiliency planning efforts involve estimating restoration times after a natural disaster. However, there are many obstacles including a lack of governmental guidance on how to estimate recovery times, an ad hoc approach of past initiatives, and widely differing disaster restoration times estimates.

How might we collect estimated disaster recovery time data from experts to use in a statistical model?

PROBLEM

Introducing Restimate: a digital expert elicitation tool

Restimate was built using Retool (a low-code platform for building internal tools) to collect data about natural disaster restoration times from emergency planners and researchers. We hope to deploy this tool for use in local emergency planning governments.

Here's my process:

SOLUTION

How 50 users estimated restoration times

I conducted a thematic analysis of user comments explaining their estimates after the first round of user tests with the MVP (conducted prior to my involvement in the project). The 50 participants were divided into groups: engineer vs non-engineer and expert vs non-expert.

The coding process:

My research advisor and I individually coded the comments from each participant. The codes were generated by identifying common trends and keywords.
We compared our tags to create a set of closed codes.
We coded the comments collaboratively again to finalize them.

Key Insights:

77% engineers had non-linear graphs while only 44% non-engineers had non-linear graphs
39% engineers mentioned topology while 11% of non-engineers mentioned it
People tend to prioritize the thing that has the highest impact. Our goal was to make sure they prioritize impact over historical case studies or rules of thumb.

From these findings, we prioritized which tags we wanted people to focus on: non-linear, topology, damage, repair effort, interdependence. These would be the main factors we would highlight through our background information and scenario descriptions. By doing so, we could help nudge non-engineers to still have estimates similar to engineers.

THEMATIC CODING

Codes for one user's comments.

Opportunities for UI improvement

I brainstormed opportunities for improving the UI of the MPV and categorized them in terms of high vs. low impact and high vs. low effort.

We focused on the low-effort high-impact opportunities:

Layout: sectioning content and using bullets
Background information - factors: explain the different terms such as redundancy and resources
Case study: show case studies without having people anchor to them
Interactivity: allow users to get view more text instead of putting everything onto one page
Scenario development: nudge considerations of factors and simplify scenario description
Maps: design with the assumption that no one will understand the maps
Instructions: be clear about whether to estimate for operability or functionality

PRIORITIZATION MATRIX

Prioritization matrix.

Revisiting the user flow

To get a sense of the path a user will take through our web app, I created a user flow diagram.

USER FLOW DIAGRAM

User flow diagram.

A blueprint for Restimate

Creating a wireframe would help us to map out the different components on each page and also lay out the content necessary for estimating restoration times for a scenario disaster.

WIREFRAMING

Screenshot of wireframe on Figma with stickies denoting design opportunities.

Building Restimate on Retool

Keeping the UI opportunities in mind as well as our insights from qualitative analysis, we developed the second prototype of Restimate on Retool.

PROTOTYPING

First page of web app describing what restoration is.

Page describing San Francisco's infrastructure systems.

Estimation page with table for user input.

Incorporating feedback and iterating

After creating this prototype, we conducted five user tests to see if we could improve the user flow and interface. Participants included three experts, one non-expert grad student, and one non-expert undergrad student.

Findings:

Users didn’t refer to the maps in their estimates so we removed them.
Users were confused about the difference between the factors for curve shape and length because the pages had the same layout. So we ended up combining them into one page.
Users had trouble navigating between pages so we added a more prominent “back” button at the bottom next to the “next” button to make it more clear.

USER TESTING

The importance of qualitative research

My biggest lesson from this project was learning how to connect with users and turning their insights into design opportunities. Without their feedback, Restimate would not have seen as many improvements and we were able to achieve our goal of creating an optimized user experience for completing a challenging task. One challenge was only having limited knowledge about natural hazards and disasters, however as a designer it’s important to remember how to gain knowledge through communication and learning to pick up new ideas quickly.

REFLECTION