This paper presents learning from the evaluation of the child poverty pathfinders in Dundee and Glasgow about evaluating evolving and complex programmes. It is intended to support policy makers, programme delivery teams and evaluators to get the best out of future evaluations of similar initiatives.