To realise a climate-neutral electricity system, residential flexibility needs to be fully unlocked. Yet, little is known about the barriers that prevent households from flexibly adapting their electricity consumption to system needs. Using advanced machine learning and causal inference tools, this innovative research project sheds a light on these barriers and how they can be overcome. Two proprietary datasets are leveraged to make this possible, for which I collaborate with different partners. First, I use data from a large-scale smart charging scheme for electric vehicles to estimate how human behaviour affects the idealised flexibility potential. Second, I analyse the load profiles of 134,000 Flemish households and estimate the causal effect (and elasticity) of introducing the so-called “capacity tariff”. Third, households owning an electric vehicle are further investigated by applying non-intrusive machine learning methods to reconstruct their charging patterns. I estimate the causal effect of the capacity tariff on charging habits, focusing on the strategies adopted by households in response to the new tariff. These three studies collectively offer the opportunity to generate innovative fundamental insights on how households interact with their flexible assets and adapt their electricity consumption in response to policy changes. By accounting for household heterogeneity and dynamic effects across time, the findings will be highly innovative and policy relevant.