There is a paucity of information regarding effectiveness of treatment-as-usual mental health services in children and young people, with uncertainties regarding which individuals improve, and whether similar outcomes would be seen without treatment. I outline my work which uses observational studies of a broad population to answer such questions, employing propensity weighting to balance the treated group and natural non-treated comparison group on confounders, similar to a randomised control trial. I discuss work in a sample of adolescents, where I showed that those without service contact had a seven-fold odds of depression three years later compared to those who had accessed services. Findings were moderated by baseline symptom severity. I extend this work in another general population sample of adolescents and emerging adults, testing whether mental health treatment was associated with change in transdiagnostic distress over two years following mental health treatment. I delve into whether effects were consistent across these two age groups, before and after the treatment “cliff edge” when adolescent mental health services terminate. Social functioning measures were assessed as additional potential treatment-related outcomes and mediators. Findings are discussed in the context of how to improve services for young people.