The COVID-19 pandemic hit many countries at a time when their education systems were facing multiple challenges. Economic, public-health and social impacts from the pandemic have exacerbated many of these challenges. The aim of this paper is to explore the way in which the behavioural sciences can help support the policy response to the COVID-induced education crisis, and to serve as a learning experience for other future crises. The paper involves an empirical exploration of the factors associated with a range of outcomes using large nationally representative datasets, and interpreting these relationships in the context of a detailed literature review. By using data that it is generally representative of the populations of interest, and is available for many dozens of countries with different histories, languages, cultures, and socioeconomic outcomes, this paper highlights how identifying behavioural biases can direct education systems towards more effective targeted policy interventions.