The concept of the Energy Poverty Observatory (EPO) aims to transcend traditional understandings of energy poverty by pioneering a federated database that will serve as a comprehensive and integrated repository, encompassing all pertinent factors impacting on energy poverty. The approach addresses previous limitations with respect to data availability and diversity. An integrated database approach aggregates diverse sources of information in near real time, encompassing surveys, interviews, historical databases, census data, sensors/IoT, and beyond. The database leverages from state-of-the-art machine learning (ML) algorithms which will be used to identify new correlations within these datasets, facilitating data-informed decision-making and policy formulation. By consolidating disparate data streams, including energy commodity fluctuations, renewable energy statistics, and socio-economic indicators, it will be possible to support holistic comprehension of the underlying risk factors and impacts of energy poverty. The efficacy of the platform will increase with time, as the ML algorithms refine their insights through the increase of historical data. Continuous integration of data will ensure the availability of timely and standardised information, bolstering the efficacy of decision-making tools and predictive simulations. The EPO is underpinned by robust theoretical framework and supported by delineated key performance indicators (KPIs) to promote reproducibility and long-term adoption by policymakers and other stakeholders alike. The EPO aspires to emerge as the reference tool for data-driven policy formulation and simulation exercises in the realm of energy poverty. By nurturing collaboration across academia, industry, policymakers, and other stakeholders, the EPO will affect enduring change in addressing energy poverty and its socio-economic implications.