What We're Learning: How to Make a Data tool Accessible and Impactful

Author:
Mark Viehman
Article

As part of our K-12 strategy, we support partners who help provide key insights on the perceptions and behaviors across K-12 education to better inform practitioners and their approaches. We know that the field has been eager to better use and analyze currently available data in order to better support the work happening in schools. The idea behind Bento, a data visualization tool for education data, came after years of manually working with output files and reports from the RAND American Educator Panels. We realized that a tool that would allow users to explore publicly available, de-identified school related data could shift the experience from cleaning files to analyzing data, noticing trends, making connections, understanding changes over time, and combining these discoveries with other information to drive decision making in support of students, educators and districts. In addition, we realized the rich data on teacher and principal perspectives could be useful and interesting to groups outside of the foundation, if available in an easily accessible tool.

 

However, acknowledging the utility of a hypothetical tool is quite different than creating something accessible, beautiful, and useful. We partnered with Kitamba, a social impact consulting and products firm, to make this into a reality. Through intensive engagement and work, the team created Bento with user needs and behavior in mind. Bento strikes a balance between being both beautifully designed and a rigorous tool for data analysis. And as user interests shift over time, Kitamba ensures that Bento also adapts and meets these new needs. This flexibility enables us and others to continue using the Bento tool for understanding data that helps us drive impact. Read on to learn more about Bento. 

-- Isabella Velásquez, Data Analyst at the Gates Foundation 

***

Putting the User First: How to Make an Education Data Tool Accessible and Impactful

Have you ever built a beautiful data dashboard that no one uses? When organizations find themselves with complex data in illegible spreadsheets, they might turn to tools such as reports and dashboards to ensure they can leverage the data for impact. However, the problem persists if these tools are unused - and more importantly, the data is still not being used to effect change.

These were some of our considerations at Kitamba as we developed Bento, a free education data visualization tool that allows users to easily visualize survey results from thousands of teachers and principals across the country and explore education outcomes data mapped by state and district. In the process of developing and scaling Bento, we learned that continuous engagement with users ensures the tool is accessible and impactful.

***

How to make a data tool accessible and impactful:

Start by solving your user’s problem:

When building a data tool, it is important to get to know your users upfront and design the tool with your users in mind.

When we initially built Bento as a pilot for State Education Agencies (SEAs) and Education Foundations, we held extensive conversations with users. These conversations allowed us to learn who they are and what is important to them and design the tool with their needs in mind. We found that our primary users were data consumers, not data analysts, looking for a tool to help them visualize complex survey data to make data-driven decisions to improve student outcomes.

 

Bento 1
After having conversations with Bento users, we created user personas to inform the development of the tool.
As a result, we designed the tool to include features that data consumers would use, and left out those that might be used only by analysts. For example, Bento users can easily filter survey visualizations, much in the same way they would filter results on a shopping website. They can also quickly export their visualizations to be used in the memos or presentations that they are building as part of their workflow.
Bento 2
For example, users can easily filter survey results by categories like teachers in California and segment by types of teachers.

As you make these types of design choices for your tool, keep a clear picture in your head of who will be using it, what their needs are, and what is important to them. This ensures you are, as the Lean Startup Methodology describes, “validating your user’s problem” and “building a product that will solve that problem” rather than executing “an idea you think people want” that might not serve their needs.

Continue to engage users while scaling the tool:

When scaling a data tool, it is critical to understand and engage users to make sure the tool is useful for a broader audience.

Bento’s initial success confirmed the potential for the tool to have a positive sustained impact in K-12 education. This led us to scale the tool so that it is free and accessible to anyone interested in using education data to inform their work. However, scaling a data tool requires significant investment of time and resources, so it is critical to understand your users' needs to ensure that the investment is not wasted.

As we scaled Bento, we needed to rebuild the tool to more flexibly ingest data and support a larger user base. At the same time, we wanted to make sure these updates were responsive to our users’ needs and goals. We did this by conducting 40 interviews with current and potential users. This allowed users to inform each step of the tool’s development, revealing new use cases and datasets of interest and shaping plans for future development. For instance, in our user interviews, we found that users sometimes struggle to find the data most meaningful to them and make sense of it given how much information is available on Bento. In response to this finding, we are adding a guided topic search that will let users find the data they care about quickly. As you expand your tool, consider engaging consistently with users to test and validate the changes you’re making to ensure they’re providing as much value as possible.

Keep users' needs in mind while solving problems along the way:

In order to balance competing priorities, it helps to understand what your primary user needs to accomplish.

When solving the thorny problems that come with building a data tool, it can be difficult to decide on the right approach. A clear understanding of users’ intentions allows you to solve the problem in a way that does not make the tool less usable. For example, Bento now displays sample size ranges for each of its survey visualizations, but originally displayed specific sample sizes. While rebuilding Bento for its public launch, we recognized that the sample sizes posed potential risks to respondent confidentiality, depending on how users filtered the data. There were a number of ways to address this confidentiality risk, including adding “noise” or masking the data, both of which would have ultimately made Bento less useful. Instead, we took into account how users were using the sample size information (for an indicator of information reliability, not for additional statistical calculations) to guide our decision. Implementing sample size ranges maintained users’ ability to understand how reliable the information is while dramatically reducing risk to respondent confidentiality. Taking into account Bento’s user’s behaviors and preferences enabled us to solve the problem in a way that didn’t lessen Bento’s value to its primary users.

 

Bento 3
Bento displays N-size ranges for all survey respondents and subgroups of interest.

Work hand-in-hand with new users to ensure they maximize their use of the tool:

Engaging with new users maximizes their use of the tool and increases the tool’s impact.

Along with engaging our current users, we found that putting in the effort to support new users is worth every minute. In all of our outreach, we offer new users a personalized Bento demonstration. These demonstrations allow us to get immediate feedback from new users we can use to inform future development. Even more importantly, they provide new users with an increased understanding of how they can tailor the tool to their needs, increasing the tool’s utility and impact. In addition, users who have had small, targeted demonstrations often come up with new applications for the tool and recommend Bento to colleagues, increasing reach.
Take the case of a non-profit focused on teacher professional development. We provided a Bento demonstration to two members of their leadership team, who then suggested we share the tool with a few more team members. As a result of this second demonstration, participants discovered additional uses for the tool and suggested it to another team. By the end of these demonstrations more than ten of their staff members across multiple teams were using the tool for purposes ranging from reporting for SEAs to identifying opportunities for organizational growth.

 ***
If you’re going to build a data tool that you want people to use, consistently learning about and responding to your users’ needs will help make the tool accessible and impactful.

This approach was successful as we built, scaled, and now continue to refine Bento, enabling  our users to access and leverage education data to drive change.

Interested in using Bento in your work? Sign up and learn more here.

 

Mark Viehman is a Principal at Kitamba. He leads their product design work, which focuses on building tools that help education organizations operate more effectively and free up educators and leaders to focus on students.