Learning Analytics are an academic field with promising usage scenarios for many educational domains. At the same time, learning analytics come with threats such as the amplification of historically grown inequalities. A range of general guidelines for more equity-focused learning analytics have been proposed but fail to provide sufficiently clear guidance for practitioners. With this paper, we attempt to address this theory-practice gap through domain-specific (physics education) refinement of the general guidelines We propose a process as a starting point for this domain-specific refinement that can be applied to other domains as well. Our point of departure is a domain-specific analysis of historically grown inequalities in order to identify the most relevant diversity categories and evaluation criteria. Through two focal points for normative decision-making, namely equity and bias, we analyze two edge cases and highlight where domain-specific refinement of general guidance is necessary. Our synthesis reveals a necessity to work towards domain-specific standards and regulations for bias analyses and to develop counter-measures against (intersectional) discrimination. Ultimately, this should lead to a stronger equity-focused practice in future.