5 data lessons we learned during the Covid-19 crisis

Eight months after the first case of Covid-19, we’re taking a look at our experiences in helping governments and non-profit organizations find, manage and share data during a global health crisis.

Data quickly became a pillar in the fight against the virus. As experts and officials worked to flatten the curve, we witnessed sharing and access to incredible amounts of data as well as equally extraordinary amounts of infographics and dashboards that circulated globally. It is as if a global resistance movement had started and it had data as its main weapon against the disease.

Sharing data resulted as an important part of the response to Covid-19. As the crisis evolved, it became evident that governments, scientific communities and civil society were in need of data-driven tools to be able to address three key objectives: (1) response, (2) reassessment and (3) recovery.

Different kinds of data can fit into each particular objective:

1. To develop efficient responses, members of the scientific community extensively shared academic research – early in the crisis, over 29,000 articles were made available by the U.S. government and partners via the COVID-19 Open Research Dataset Challenge;
2. To allow governments to identify trends and reassess quarantine measures, mobile companies, such as SFR, and tech giants, like Google and Facebook, published anonymized geolocation data from their users.
3. To help countries recover and relaunch economic activities, public institutions promoted local tourism – our client Ile-de-France created the initiative “My summer, my region” to help citizens find holiday plans in their local area.

With a leading data-sharing solution, Opendatasoft provided our tools and best practices to clients and pro bono partners to help them achieve these key objectives.

We created a website to showcase our clients’ initiatives, such as sharing locations of open pharmacies or food and beverage businesses during the response phase, and information about government-led economic aid programs during the recovery phase.

Additionally, as an ongoing reassessment measure, our team built and shared the code for two Covid-19 applications: a dashboard to illustrate key health indicators for Covid-19 per country, and a dashboard template that could be customized to match region-specific needs.

As part of the Customer Services team, I was responsible for supporting clients and partners as they gathered relevant data and deployed dashboards. I had the opportunity to work closely with public institutions in France, Spain, Switzerland, the U.S. and Mexico as they searched for ways to respond rapidly and responsibly to the crisis, providing citizens, civil society, industries and public officials with the necessary knowledge to act.

I experienced first-hand how data sharing can be a powerful weapon in enabling governments to provide emergency response, but also how any response based on data can be undermined if the data is unreliable. Poor data quality can lead to blind spots and biases in health-related data, which during this crisis has usually been a result of under-testing, government misreporting or lack of proper expertise when exposing data.

Thus, how exactly can government institutions better prepare to develop data-driven rapid and efficient responses to a crisis? What pre-conditions favor it and what common challenges did we encounter as we worked with different regions in the world?

There’s no doubt that having data related to a global health crisis will have a significant impact on your website’s traffic. With data sets updated daily,  sometimes even more than twice a day, the influx of API calls and downloads in Opendatasoft portals sometimes increased by several orders of magnitude when compared with previous averages and surges – some portals reached over 1000 times their usual traffic!

Having a solid, trustworthy and scalable infrastructure capable of taking on this overcharge is crucial. Here at Opendatasoft we are firm believers that cloud-based solutions (such as ours ) are the way to go when looking for these elements.  

Cloud solutions are reliable because they have built-in high availability and easy failover. A cache system can, for instance, allow the software’s infrastructure to easily absorb a traffic peak by saving similar requests and simply reproducing them as new ones arrive. When chosen correctly, they are also trustworthy for showing high levels of regulatory compliance.

Last but not least, cloud-based solutions guarantee scalability when compared to hardware-based ones.  The Covid-19 crisis illustrates the advantage of using the cloud, as it can easily increase its capacity to absorb intense peaks of traffic within minutes. On the other hand, using hardware requires buying and setting up additional servers, which could take weeks.

From how data is being collected to the way it is processed and then publicly displayed, there are several steps and people involved that may undermine its quality and slow down an organization’s ability to share it.

In times of crisis things need to get done fast. Imagine you work for a city and you would like to build a Covid-19 monitoring dashboard. Where would you collect this data from? Who would you need to talk to and how would you make sure the information you are sharing is reliable and respectful in terms of privacy?

When a crisis comes, organizations should be well-equipped to approach new data challenges because they already have established mutual trust,  processes, and ambassadors to help the rest of the organization move seamlessly.

Our client, Junta Castilla y León in Spain, provides a good case study. Only a couple of weeks after the pandemic hit Spain, the region had already published a dashboard to share real-time health information with citizens and public officials. They constantly improved the dashboard by adding new data and analysis as the crisis evolved (the number of people unemployed and receiving government aid, the region’s testing capacity, number of patients moved to other areas, available medical personnel and the percentage of those who had contracted the disease, etc).

If you have a few minutes after reading this, you should go check out their dashboard – it is quite amazing!

When asked about what key success factors they could share with their peers, their Chief Transparency Officer,  Antonio Ibáñez Pascual, gave 3 main ones:

1. Build a data-driven culture while collaborating with both internal and external stakeholders. At the Junta de Castilla and Leon, this is driven by top executives. Before the Covid-19 crisis, the transparency team had already built strong relations and worked hand in hand with the region’s health department. They were familiarized with the way health-data was collected and knew who to contact and which steps to take in order to publish this data in their portal.
2. Be reactive to feedback. From journalists, to researchers or the average citizen looking to stay informed on the Covid-19 numbers for the region, the messages they received on a daily basis were thoroughly read and helped them improve data visualizations or look for more data that could enrich the analysis.
3. Be as prepared as you can be. While building strong ties with other departments and iterating with the community, the Junta of Castilla and Leon has published more than 200 datasets since the launch of their open data portal with Opendatasoft in 2019. They have built several dashboards such as this one on energy consumption, allowing their team to go through the necessary training and knowledge-gathering to deploy one for Covid-19  with such short notice.

One common (yet not new) challenge observed during the Covid-19 crisis is the lack of data quality. 

As an organization that is sharing data with the goal of informing emergency responses, reassessment and recovery initiatives, you need to provide essential information on how this data was collected, how often and where. In other words, you need to fill in your metadata.

Metadata is a topic that we care deeply about here at Opendatasoft. We’ve published several guidelines and best practices – check out our most recent white paper on metadata.

Why is metadata so important? Think of it this way: you work for a city council and you need to inform public officials on the population’s health conditions to help them reassess quarantine measures. You obtain raw data shared at the national level but there’s no information on its date of modification or update frequency. You might have today’s numbers but if you can’t certify it, the information you share with public officials will be irrelevant.

Providing metadata also means that you have to respect standards. Whether these are metadata standards, such as DCAT for Europe or specific methods and guidelines when collecting and sharing data, standards are what allow systems to become interoperable, and exchange and interpret shared data.

Interoperability is crucial in providing experts with consistent indicators, especially during a global crisis, that enable them to give policy recommendations across different administrative levels. For instance, one key step when it comes to interoperability is being able to download data in non-proprietary standard formats with clear documentation.

These elements require strong cooperation between entities that will collect the data and those that will reuse it.

Our client, Mexico City, has been doing a remarkable job improving the quality of their data sets by working on the city’s standard metadata guidelines. They applied these guidelines during the Covid-19 crisis by asking our team to add custom metadata to their portal, such as “data source”, and providing users with a dictionary of the data exposed.

In a recent discussion with their innovation team, Alejandra Gonzalez, Head of Open Government, described Opendatasoft as “a real-life sandbox for creating the city’s data management policy”. She described how publishing data in their open data portal had helped them interact with their community of users and understand their needs.

Many of these users came from government agencies that were also sharing data with the city’s innovation team and wanted to obtain information from other entities in return. The standard metadata guidelines spurred the creation of a participatory process that established the city’s data governance policy.

As the crisis evolved, so did the number of health monitoring dashboards you could find. Everybody was counting on a daily basis the number of cases and deaths in each part of the world as if we were counting medals during the Olympics. When looking at Covid-19 health data from Peru or Italy, without context, you might as well be comparing apples and oranges. 

For example, the famous “curve” infographic that shows the cumulative number of cases of each country needs to be contextualized with testing capacity and criteria so that the data is accurately interpreted.

While some countries or regions were testing cases with mild symptoms, others were only counting those that had been hospitalized. The organization Our World in Data clearly depicts these nuances.

As you can see on the graph, each country’s curve is color-coded to show whether or not they are testing adequately. This is measured by the rate at which a country is finding a positive case for every few tests performed. Further information on this methodology can be found here.

Giving context to your data is not only important for comparisons at the international level but also at the national level. Data aggregations can improve reliability but obscure trends at the local level. It’s crucial to allow government, analysts and civil society organizations to see how different geographic areas and certain communities are being affected in order to provide tailored responses.

The city of Newark in the U.S. applied this best practice by showcasing aggregated data at different administrative levels: state, county, local. As one of our pro bono partners during this period, their dashboard highlights discrepancies between Covid-19 related data based on the population’s demographic and geographic distribution.

While some of the previous takeaways explored in this article highlight challenges to overcome, this last one indicates that we are on the right path. 

Remember our dashboard to illustrate Covid-19 key health indicators per country? Well, this was only possible thanks to the amazing work of our data team who collected data from official government sources, such as Santé Publique France, and shared it with our community through Opendatasoft’s data network.

Many of our clients, such as the region Centre-Val de Loire in France or the city of Basel in Switzerland were able to inform citizens about Covid-19 health indicators thanks to their work, which was only possible thanks to the data shared by public health institutions. 

This initial part of the chain for data sharing shows the potential of such practice when it comes to solving public issues. Had Santé Publique France not provided reliable and up-to-date information on the Covid-19 pandemic, citizens and public officials would have been kept in the dark and lacked the proper tools to fight it. 

Local communities also played an important role to enrich the information that was being shared with the public. As part of our pro-bono partners, we worked with the NGO Cambridge Local First in Boston, Massachusetts to help people from the region find products and services they needed while quarantine measures were being applied. With a mission to support, promote, and celebrate a “local economy community,” the organization used Opendatasoft to crowdsource data from local business owners, and display it in a user-friendly format for citizens.

Although the takeaways listed here may not be new, they have shined a light on best practices for fostering data-driven policy making. From having a robust infrastructure to respecting standards and improving data quality, our experience working with clients and partners during the pandemic showed us how, combined, these aspects can lead to more accountable and efficient governments, foster community collaboration and knowledge-gathering to better tackle public issues. 

During these times, we’re witnessing how the power of data sharing can enhance our society’s ability to be resilient. Although there are still challenges public institutions need to overcome to achieve best practices, this crisis has reaffirmed the great potential and advantages sharing reliable and relevant data.

Here at Opendatasoft we are proud to have supported our clients during these difficult times, and will continue to provide our software and expertise as the situation evolves. A sneak peak into a project our Customer Services team is currently working on is this dataset that will gather all of our clients’ initiatives around data sharing.

Our goal is to constantly inspire and promote successful use cases throughout our community of users, delivering tailored recommendations according to their specific needs. We’ll never get tired of saying that sharing is indeed caring.