Realist research is increasingly used to evaluate complex interventions. However, it can be challenging to codify and implement, with few examples to guide the process. This article describes how a team of social care leaders, practitioners and researchers developed initial program theories for the Trauma Consultancy Service (TraCS) in early learning settings, as the first phase of a realist evaluation. It explores conceptualisation of realist terminology, design and facilitation of realist interviews, and data coding and analysis using retroductive reasoning. Qualitative interviews with the TraCS team focused on understanding contextual factors, resources provided by TraCS consultants, changes in educators’ reasoning and how components interacted to generate educator and child outcomes. Eight program theories capture how TraCS supports educators to develop a trauma-informed lens and practice. This research contributes to understanding of the benefits of welfare sector-driven consultancy in early childhood.