Trust, humanity, equity, and privacy are the four pillars of the responsible use of data. These themes emerged spontaneously from presentations given at Data on Purpose: The Promises and Pitfalls of the Connected World, hosted by the Stanford Social Innovation Review (SSIR) in February 2018. The conference was about the current state of data for nonprofits, and a glimpse into what the future holds.
Several best practices for designing big data systems, measuring impact and trends, and making decisions about what’s next were also revealed. Taken as a whole, they form a guide to how we should move forward into a data-driven landscape.
The current state of data: messy
SSIR Managing Editor Eric Nee said less than 0.5% of all data was ever analyzed and used. While surprisingly low, it makes sense, given that the rate at which we’re collecting data is exploding. When a field is growing so fast, it’s a challenge to build streamlined, polished systems that can keep up. Often, there are no common standards for data formatting. This makes it challenging to aggregate data and to obtain more informed insights.
People measure the wrong thing. Erika Salomon, a data scientist at the University of Chicago who worked on the Data Maturity Framework, said the desired social change outcomes are often not captured within data sets. For example, a homeless organization might count the number of people in beds at a shelter every night, but that doesn’t tell us whether people’s lives are actually improving.
Nonprofits may lack internal data culture. Data and tech exist in silos, and may not be included in decision-making, reporting, and planning processes. Or when data is collected, people don’t know that proper formatting and consistency are essential to making sense of the inputs.
The future state of data: exciting and unknown
Moving forward, our ability to use data to make good decisions will become dramatically more powerful. We’ll have sharper insights, more quickly, and more reliably than ever before. We will harvest vast amounts of ever more fine-grained data. It will become more readily available as the world moves to a more data-transparent, query-able environment. This will happen alongside the increasing adoption of the internet of things, wearable devices, and the current lack of regulation on access and use of data.
We’ll be able to make accurate predictions. The rise of AI means much faster collection and automation of analysis processes. With enough fine-grained data, we can build predictive data systems. For example, migrating birds may need pop-up water spots in drought-ridden areas at certain times, and the ability to forecast that means we have the opportunity provide a solution.
The four pillars
1. Trust is key
It’s critical that people trust organizations using data, the data itself, the algorithms used to parse it, and the assumptions and biases built into the methods of collection and analysis. Those who control data must be good stewards of that power.
Organizations must ensure how and why they are collecting data are mission-driven practices. One method is to “champion the customer,” according to Erika Salomon. Organizations should listen, and employ participatory design practices to achieve positive outcomes.
Communities and individuals must have input. Big data must be given local context in the form of “ground-truth”—information coming directly from the people, place, or things surveyed. For example, the Streetwyze platform, which uses community insights about neighborhood health to empower local residents. Without the local context big data is at risk of providing inaccurate insights, which can lead to detrimental and discriminatory policy-making and power structures.
Trusting algorithms really means trusting those who wrote them. Data practitioners should beware of bias built-in to algorithms that collect and analyze data, and should examine the assumptions that may be embedded in the methods of data collection and use of data.
2. Remember your humanity
There is fear in the nonprofit sector that data turns human beings into cold, hard numbers. Finding ways to humanize data and even how we talk about data is critical to widespread acceptance of data-driven culture. It’s a basic human need to want to connect with others. We must keep this in mind, and ensure data systems advance humanity.
Data is simply information. Technology is simply a set of tools to manage information. Both information (data) and tools (technology) can be used for social good. This kind of straightforward language can help open doors for conversations about the benefits of data, and change perceptions of its value.
Pursuit of efficiency can result in dehumanizing processes. At Laboratoria, a web development and job skills service for Latin American women, they initially posted just the technical bios of women seeking work in coding jobs. But when they included personal stories about the women, the rate of hire increased dramatically. Hiring managers said it improved the interview process because they felt like they already knew the candidates.
3. Community involvement drives equity
Equity must drive the construction of new data ecosystems, methods, and tools. The concept of “ground truth” is part of this equation—ensuring that local communities have power over their own data. But there’s more to consider.
There are several obstacles to data equity that must be considered in development of any data practice, which include:
- Availability: Can you get to it?
- Access: Can you actually use it?
- Awareness: Do you know that it exists?
- Affordability: Can you find it, access it, or afford it?
- Agency: Are you able to use it the way you want?
- Ability: Do you have the know-how to get it done?
Inclusiveness and participatory design are two ways to move towards equity. Organizations should involve the surveyed communities, or better, let them lead. People and communities should own their own data and systems. Practitioners should learn to balance their power, and scale back their role—leaving room to empower others.
4. Privacy is everyone’s responsibility
It’s important that people and communities are in charge of their own data. They should be able to have a say in, if not outright control of, what happens with data collected from them.
There is risk of abuse when an organization captures large amounts of detailed data. Planet, Inc. has thousands of cameras and satellites orbiting the earth, snapping photos of every square meter of the globe. Andrew Zolli, VP, promised us that the company would never give the data to anyone with nefarious intent. He was convincing. But there’s really nobody to stop them. We could never really know the level of detail the company is actually collecting. We must take Zolli at his word. The recent Facebook/Cambridge Analytica scandal underscores this problem.
People should control their data the same as they control their money. Equitable legislation (and enforcement thereof) that dictates how corporations may use an individual’s personal data would help. Mark Zuckerberg’s testimony about the Facebook platform’s use of data is highlighting the importance of this. The EU’s recent release of the General Data Protection Regulation (GDPR) leads the way on equitable legislation, and serves as a model for other parts of the world.
Nonprofits need a data culture
75% of NPs collect data but only 6% think they use it well, says Stanford Program on Social Entrepreneurship lecturer Kathleen Janus. By growing a strong data culture, nonprofits can gain real insights about the realities, scale, and impact of their work, and have the information to adapt when necessary.
Data and technology should flow through all aspects of operations and programming. Staff should be trained on why quality data collection matters, and how to do it. The key is remembering the humanization factor. Ultimately, it’s not about the data. It’s about people.
Even small organizations can and should build a data culture. They should focus on what they can realistically measure. It doesn’t need to be sophisticated. It can be as simple as tally marks on a whiteboard or a basic spreadsheet. They could pick 3 to 5 key metrics or even measure proxies if direct measurement is out of reach.
Sharing how and why organizations are collecting data is becoming a donor requirement. To meet this requirement means organizations must first build a strong data foundation. It also requires funders that want the data to help pay for building the culture, and collection, analysis, and reporting.
Connect data sets for stronger insights
The lack of data standards across most sectors makes connecting data sets difficult. But it won’t be difficult forever. By doing this, we gain new insights, identify new patterns or trends, and see systems from a different vantage point. All of this helps us build tools that benefit the sector as a whole, and help everyone to make more informed, better decisions.
In 2012, the US Internal Revenue Service made Form 990 data open source. It’s an underutilized and incomplete data set, but it’s still possible to build a data engine that could parse it in various ways to help us better understand the ecosystem of all environmental nonprofits, for example. Charity Navigator has, in fact, created a repository for anyone to use to explore the 990 data.
A focus on building tools that can help entire communities of practice, systems, or sectors can drive the process of connecting data sets . And, if we keep in mind the need for standard formatting across individual data sets, the connection process can become easier as we go along. This can empower people to solve problems faster and more effectively.
Measure (and share) your impact
Once you’ve built a humanized, trustworthy, equitable data system that’s collecting and aggregating data, it’s critical to know if your efforts are working. Measuring impact lets you tell and prove your story of impact to funders. And it provides internal information necessary to make improvements to operations and processes.
There’s more to proving impact than having data. Weaving your data into a good story is a powerful way to capture emotion while providing evidence of your success. Choosing the right data and story is essential.
In choosing what to measure, distinguish between outputs and outcomes. Outputs are vanity metrics, and are often easy to measure. Outcomes are indicators of actual behavioral change, but can be harder to measure.
Dr JaNay Nazaire, Managing Director for Performance and Results for social change organization Living Cities, provided a framework for data-driven decision making. It can help organizations identify the right metrics. These are four of the five steps.
- Define the problem: What are the root causes? What are measures of success?
- Take a data inventory: What do you have, need, want? Is it accessible? Does it need work?
- Be smart about data collection: What can you start with? How can you fill gaps? Who can help you?
- Communicate clearly and powerfully: What can everyone understand? What will galvanize people? What visuals will get attention?
In answering these questions, organizations can discover the metrics that prove they are actually solving problems. Your critical metrics must address root problems and measures of success, be something you can actually measure, be realistic in terms of your ability to measure it, and support your powerful story of impact.
Better decisions with data
Data-driven decision-making is a way to systematize and validate the critical decisions nonprofits must make. Start by identifying key metrics. Then set up a system to collect data about them, and frame a powerful story of impact supported by them. Now you’re ready for the fifth step of Dr Nazaire’s data-driven decision-making framework. This is where you make the actual decisions.
Business intelligence (BI) tools can offer powerful insights. Jaclyn Roshan, from myAgro, discussed their use of BI. There’s a data warehouse, data mining, analytics, a dashboard, and data visualization of outcomes and processes. This helps them make smart, timely decisions about the farmers they serve. There was a case where many crops had become infected with a disease early in the harvesting cycle. But through their business intelligence they were able to tell farmers the best time to harvest to get the maximum crop yield before the disease destroyed the entire crop.
Cross-referencing different data sets yields insights. Impact View Philadelphia uses the 990 data and data from the US Census Bureau to display information about local nonprofits juxtaposed with information about the people living there. You can see where organizations are on the map, what type of organization it is, and add demographic layers like median income or poverty rate to understand proximity of organizations to those in need.
Access to data sets is expanding all the time. By learning more about ourselves and our communities, and by connecting data sets in smart ways, we can start to build data-driven decision-making tools that lead to improvements in people’s lives and communities.
Trust, humanity, equity, and privacy should be at the heart of our data-driven future. What we build should be in service to the communities they are for. Data systems should be protected from potential abuse through legislative changes, changes in business practices, and education of individuals.
Organizations should begin building a data culture and start sharing data. This will allow them to measure and prove their impact for fundraising and self-improvement. It will also help them make informed, data-driven decisions.
If we handle our emerging ability to gather, analyze, and share massive amounts of data well, we will be poised to make truly dramatic improvements to how we handle all the systemic problems nonprofits exist to solve. Ultimately, it could result in a world that is more equitable, supportive, and safe for all.