Designing for communities in crisis
How we designed the next phase of Safety Check
Nothing matters more to people than their safety, especially during a crisis. We built Safety Check a few years ago to make it easier for people to let their loved ones know they’re safe after a disaster and check on friends in the area. Even though communities have activated Safety Check hundreds of times across the world, we know Facebook can be doing more to help empower and connect these communities during a crisis.
As a product designer on the Social Good team at Facebook, I have the amazing opportunity to build crisis response tools for over 1.86 billion people across the world. Designing for such a large, diverse audience is an intimidating task because these tools need to work across a number of factors like languages, cultures, borders, network conditions and so on. Our team in particular faces the added challenge of ensuring that the tools we design can be used in difficult crisis environments when people are under extreme duress.
How do you design for a billion people?
To build tools for such a diverse audience, our process begins with one word: empathy. That means really knowing who we’re designing for and understanding the problems they face. At Facebook, we care deeply about this so we practice it in two ways:
1) By using qualitative insights from user research where we talk to people learn about their experiences and needs. We like to use a method that researchers call ‘contextual inquiry’ where we go to people and observe them in the environments where they use our products.
2) By using quantitative data to identify patterns in the usage of existing solutions and potential opportunities for us to improve them.
Identifying the problem
With the goal of understanding our audience and designing with empathy, we travelled across the world last year to communities that had experienced different types of crises. In early 2016, my team and I were in Chennai, India to learn more about people’s experience with the devastating floods they had recently experienced, whether Safety Check was useful to them and what else we could be doing to help. During this week, we heard story after story of people opening up their homes to their neighbors, cooking and traveling miles (sometimes swimming through chest deep water) to deliver food to badly affected neighborhoods, and even risking their lives to save stranded animals. As we travelled to other cities around the world, we quickly realized that these powerful stories of generosity were not unique to Chennai. We heard similar stories in Kumamoto after the earthquake, in Paris after the terror attacks, in Fort McMurray after the wildfire and in Oakland after the warehouse fire. It became clear to us that regardless of borders, cultures, languages, or the type of crisis, there were two insights in common across these places:
1) Communities are resilient. There was an overwhelming sense of goodwill and kindness in people after the crisis and they really came together to support each other in times of need.
2) Even though Safety Check worked exactly as intended and was a valuable tool to quickly reassure loved ones of your safety, people in the affected area had a lot of other needs like relief supplies and services, access to information, etc., and there was a lot Facebook could be doing to help. The need for basic supplies and services like food, water, shelter, clothing, and transportation are higher and more exaggerated during natural disasters that disrupt the infrastructure of the affected area, especially in the short term after a crisis.
Our research showed that people go out of their way to help others in their community with needs like food, water, shelter and other basic necessities.
We saw these insights reflected quantitatively in data as well.
These communities were gathering on Facebook Groups, using public spreadsheets or ad-hoc sites to organize, and social media to communicate. In addition to using these tools to share information, ask questions and get updates, they were using it to connect people who needed help with those who could offer help. While these were great uses of existing tools, they weren’t very successful at creating those connections and very often needs went unmet. Since these solutions were not built for this specific scenario, there were a few key problems that people faced:
- The solutions were fragmented. People were using a variety of tools, from social media to public spreadsheets, to organize and they didn’t reach the whole community. It was hard to find the right forum so people would often have to post in multiple places.
- The solutions weren’t searchable. When people were looking for help, there was no easy way to find what they were looking for without scrolling through hundreds of posts. There was no easy way to match with others so in a lot of cases, needs went unmet.
- The solutions required manual curation and maintenance. We heard from volunteers on the ground that they would mine posts from social media and plug them into spreadsheets or plot them on maps to try to connect people who could help each other. This made the matching process inefficient and slow. They even mentioned instances of duplicated efforts because there wasn’t an easy way to know if the person still needed help, if supplies were still available, and so on.
Designing the solution
We wanted to solve some of these fundamental problems by designing a tool that would connect people who needed help with those who could offer help in the community more seamlessly and efficiently. We believe that Facebook is the right platform to do this because this is where communities naturally gather currently and we can create connections between people better than anyone else. So starting from sketches and paper prototypes, we started to design and iterate towards a solution.
In the spirit of identifying existing behaviors and building upon them, we analyzed hundreds of sample posts on Facebook Groups where people were organically requesting or offering help after a crisis and scanned them for similarities. What types of things are they requesting for or offering? Do they always provide contact information? Do they always specify the number of people the help is for? What other critical information do they share? How do people respond to these requests? We then experimented with turning some of this commonly shared information into structured fields while composing a message.
During UX research, we would do exercises with participants who have experienced crisis situations to break down the post-crisis timeline into phases and list specific needs (both physical and emotional) that arose in each phase. What happened in the minutes/hours/days/months after a crisis? What did they need? How were they feeling? Insights from these exercises and consulting with humanitarian relief organizations helped us determine what categories represented the most critical needs during natural disasters.
These are just two examples of how we started identifying patterns and differences in existing behavior to leverage in our designs. Over the course of 3–4 months, we tried several other such approaches and iterated continuously while learning from field research, usability testing, industry experts, and humanitarian relief organizations. It was important for us to be constantly checking-in with both users and experts to make sure that the solution we were iterating towards was always solving the right/most relevant problems.
After months of hard work and collaborating cross-functionally with a variety of teams, we arrived at our solution: a new feature called Community Help that lives within Safety Check. This means that in addition to telling friends you’re safe and checking on them, you can also find help or give help to people in your community. Categories for help include things like food, water, shelter, transportation, baby supplies, equipment, toiletries, and a variety of other relief supplies and services that we learned are important for people in the recovery period after a crisis. Community Help improves upon existing tools that people are using in several ways:
- Integrated in a single place: Since the feature is within Safety Check, it means all the people that are coming to Safety Check through notifications and friend invites can now access this feature without having to create separate groups or documents.
- Structured data: The posts made through Community Help are searchable and discoverable because we’ve added structured data like categories, number of people, and an approximate location, which will help create many more successful matches between people requesting help and offering help than before.
- Location-centric: Since each post has an approximate location associated with it, you can see all the offers and requests for help plotted on a map to easily gauge how far away from you they are, which is really important during a crisis since accessibility is often impacted during natural disasters.
- Fresh feed: In addition to the posts being ordered in reverse chronological order, people have the ability to close out their requests or offers for help, which helps keep the Community Help feed fresh and updated.
- Contacting people made easy: All posts have a ‘message’ button that opens a private thread on Messenger or you can comment on the post.
It’s really exciting that after nearly a year since I first sketched this idea in my notebook, it has finally come to life and is officially starting to roll out today. To say that this launch has been a huge team effort would be an understatement. Each and every member of this team has an incredible passion to do good and create positive impact in the world and it has been really inspiring to work with them. We’re just at the start of this journey though and we’re going to keep working on making these tools better and more useful for communities in times of need.