The standard approach to answering a causal query (e.g., P(Y|do(X)) when given a causal diagram and observational data is to generate an estimand, which is an expression over the observable variables, if the query can be answered uniquely. The estimand is then evaluated from the observational data. In this paper, we propose an alternative paradigm for answering causal queries. We suggest learning the full causal model from the observational data given the diagram. Once a full model is available, Probabilistic Graphical Models (PGM) algorithms developed over the past three decades can be applied to answer the query. We present this idea and provide analysis, demonstrating that this approach can be far more effective than the estimand-based approach with plug-in estimation, when the diagram has a low induced-width. Our analysis and experiments illustrate the potential of this approach over a collection of synthetically generated causal models.