Building the Right Thing: How Product Thinking Guides Our Work at OpenSAFELY Schools

What is product management?

At OpenSAFELY Schools, we use an iterative product development approach to build software that solves real user problems, aligns with organisational goals, and is technically feasible. This means we don’t just jump to a solution — we explore the problem space first, test assumptions, and build incrementally.

Our approach combines design thinking and lean startup principles. We start with user research to understand pain points, align with stakeholders on goals, and then build prototypes of potential solutions. These are tested with users, and based on feedback, we adapt. We continuously repeat this build → measure → learn cycle to ensure we’re building the right thing — not just building things right.

Understanding the problem

Our first product goal was to create a Trusted Research Environment (TRE)-like system where education researchers could securely access a newly created anonymised dataset, known as TED (Teacher Education Dataset).

To understand how researchers currently work — and where they face friction — we began with user research. We spoke to TIDE (Teaching Improvement through Data and Evaluation) researchers (the research that uses data from TED) and mapped out their journey and pain points. We also spoke to potential new users, such as education economists and data leads at Multi-Academy Trusts (MATs), to understand what questions they wanted to explore with TED that they couldn’t currently answer.

The most common and urgent pain point? The time it takes to prepare the data.Currently, it takes about two months per MAT to get the data research-ready. The causes are many: data is stored differently across MATs, cleaning takes time, and data requirements vary year on year. This severely slows down the research progress.

Testing our hypothesis

Based on this, we formed a hypothesis:

If we adapt ehrQL — a human-readable query language originally developed for healthcare data in OpenSAFELY — for use with education data, we can reduce the time researchers spend preparing data and make querying TED more accessible.

We tested this idea by building a prototype using a modified version of ehrQL using education data. We created mock queries to simulate the experience and shared it with a small group of users. The initial feedback was encouraging — users found the queries more readable compared to what they currently use.

Building and testing the solution

We are now developing an early working version of education-focused ehrQL. This will allow researchers to query TED directly, without needing months of manual cleaning.

Next, we’ve lined up user testing sessions to validate our latest hypothesis:

That ehrQL is readable, understandable, and usable by education researchers.

If successful, this will shorten the time it takes to prepare data, broaden access to research-ready data, and support a wider range of users engaging with TED.

What’s next?

By involving users early and often, and aligning closely with our stakeholders, we’re building a product that truly meets the needs of education researchers. This approach doesn’t just help us build better software — it builds trust, encourages engagement, and ensures our work has real impact.

We’re excited to see how these early steps with ehrQL and TED evolve, and we’ll continue sharing what we learn along the way.