The science comes first. The product follows.

Socratic Questioning

The oldest teaching method is still the best.

The Socratic method — teaching through questions instead of answers — has been around for 2,400 years. Across the research we analysed, one recommendation came through more consistently than any other: teach through questions, not answers.

The reason is simple: when someone asks you the right question at the right moment, you have to think. Not recognise. Not recall. Actually reason through the problem. That's when understanding happens.

Active Learning

Students who think, not watch, fail at half the rate.

A landmark meta-analysis of 225 STEM teaching studies (Freeman et al., 2014), published in the Proceedings of the National Academy of Sciences, found that students who actively work through material fail at half the rate of those who passively receive it.

You can watch an excellent explanation and feel like you understand. But feeling isn't the same as understanding. The test is whether you can explain it back, without looking.

Retrieval Practice

Testing yourself beats re-reading. By a wide margin.

This is one of the most replicated findings in learning science. A major meta-analysis (Adesope et al., 2017) covering 272 studies found that retrieval practice — the act of trying to recall something from memory — consistently outperforms re-reading, re-watching, and every other passive study method.

The counterintuitive part: the effort of trying to recall something — even when you get it wrong — strengthens memory more than re-reading the answer ever could. Re-reading feels productive. Retrieval practice is productive.

Spaced Repetition

Spreading study over time beats last-minute cramming.

When spaced practice is combined with retrieval practice, the combined effect on learning is among the strongest measured in educational research.

Your brain consolidates knowledge between sessions, not during them. Cramming everything into one night creates the illusion of readiness. Spacing creates the reality.

Personalised Adaptive Learning

One-size-fits-all doesn't fit anyone.

A meta-analysis of 34 studies (Zheng et al., 2022) found that adjusting content, difficulty, and pace to individual learners produces significantly better results than generic delivery.

There's an interesting nuance in this research: students learn significantly more from adaptive instruction, but they don't always feel like they are. The effort feels harder, because the difficulty keeps pushing at the edge of what you can do. That discomfort is the learning.

Self-Explanation

Explaining why something works produces deeper learning than being told.

Across 64 studies (Bisra et al., 2018), a clear pattern emerges: when learners are prompted to explain reasoning — not just state facts — comprehension deepens significantly. Open-ended explanation outperforms multiple-choice and all forms of passive instruction.

This is the scientific mechanism behind Socratic questioning. When you're asked "why does that happen?" you're not just retrieving a fact — you're constructing understanding.

Formative Assessment

Checking your understanding as you go beats a test at the end.

An umbrella review of 13 meta-analyses (Sortwell et al., 2024) covering more than 256,000 students found that ongoing low-stakes assessment during learning — not just end-of-unit tests — produces significantly better outcomes. And the biggest gains come when students check their own understanding, rather than waiting for a teacher to test them.