Formal verification relies on specifications formalized in Computational Tree Logic to express desired system behavior. In practice, specification suites grow and evolve during iterative design cycles, accumulating overlapping and semantically dependent properties that increase engineering and verification cost. We propose to address this \emph{specification debt} via \emph{property refinement}, which captures semantic relationships between properties, identifying when one property is already guaranteed by another. This provides engineering insights into these semantic relationships, easing requirements management and enabling the identification of redundant properties. Our preliminary evaluation shows reductions of up to 70% and speedups of up to 30 times, showcasing the practical potential in optimising verification workflows and guiding engineers towards identifying what is really needed.