Selected work

Learning Commons · Concept testing

What developers need to know before they commit to a dataset

Concept testing the first version of the Learning Commons dataset catalog

Company

Chan Zuckerberg Initiative’s education initiative, offering open education datasets for developers

Role

Senior UX Researcher, Learning Commons

Partners

Developer experience PM, two dataset PMs, design, partnerships, trust

Methods

Concept testing with an interactive prototype, moderated sessions

Timeline

3 weeks, 2026

In short

Learning Commons was getting ready to launch a public catalog of education datasets for developers. As the only researcher, I tested an interactive prototype with 10+ developers at edtech companies to define what the first version needed. The concept worked, but dataset pages didn’t tell developers enough to judge whether a dataset fit their use case. The team rewrote the pages, made the most common access option the first action, and planned the next version around exploring data before committing to it.

The problem

Learning Commons offers open education datasets that developers can build on. Before the public launch, the team needed to know whether a catalog of those datasets made sense to developers, and what the first version had to get right. Three PMs, design, partnerships and trust all had a stake in the answer.

What I did, and why

1. Set four questions. Did developers understand the catalog overall? Did they understand the ways they could access each dataset? Was the information on each dataset enough to judge it? Could they learn and navigate the catalog on their own?

2. Recruited the people who’d use it. I recruited 10+ developers at edtech companies.

3. Tested a working prototype. In 30-minute moderated sessions, developers explored an interactive prototype as if they were looking for data for their own product.

4. Ran it end to end in three weeks. I owned planning, recruitment, moderation and synthesis.

What we found

  • The information on each dataset wasn’t enough. This was the biggest gap. Developers couldn’t tell from a dataset page whether it would work for their use case.

  • Navigation worked, but the design rushed people. Developers found their way around easily, but the interface pushed them to take action before they understood the dataset.

  • The concept held up. Developers understood what the catalog was for, which confirmed the team’s direction.

What changed


The research set how the catalog is organized.Dataset pages got clearer, shorter descriptions of each dataset and its use cases.The first action became the most common access option, so developers could explore and get started right away.The next version’s roadmap focuses on helping developers test and explore a dataset before they commit to it.

What I learned

  • Technical users decide fast. Developers wanted to get a sense of the shape of the data, then get something running to test before committing. If we didn’t answer those questions quickly, they’d leave. That’s why the first version leads with clear descriptions and the most common way to get started.

  • The right answer has to be one the team can ship. What we offered also had to fit what our product teams could deliver and what partnerships and trust allowed. Part of my job was finding the first version that met developers’ needs inside those limits.

Next case study

Why teachers weren’t using a product their schools paid for

TPT · Journey mapping